From syounger at boulder.net Sun Sep 2 22:19:22 2001 From: syounger at boulder.net (A. Steven Younger) Date: Sun, 2 Sep 2001 19:19:22 -0700 Subject: meta-learning Message-ID: <003b01c1341e$db06be40$922b97cc@syounger> Dear Connectionists, In a connectionist message on Aug 30, 2001, Juergen Schmidhuber graciously drew attention to some work on meta-learning (learning how to learn) done by Sepp Hochreiter and myself. This work was presented by Sepp at the ICANN conference in August of this year [5]. For those of you that are interested, we have a companion paper that I presented at the IJCNN conference in July of this year [9]. This paper describes our work from a different perspective - that of "Fixed-Weight Learning." Some other related papers are: L. A. Feldkamp, et al. in 1996 [3] used a similar Fixed-Weight Learning method to meta-learn time series prediction problems. The possibility of this type of meta-learning was conjectured by N. E. Cotter and P. R. Conwell in their 1990 and 1991 [1,2] original papers on Fixed-Weight Learning. Some examples of neural networks of this type were presented by myself, Conwell and Cotter in a 1999 paper [8]. Another type of neural network that has embedded learning was presented by Schmidhuber in 1993 [6]. An explanation of the Long Short-Term Memory (LSTM) was presented by Hochreiter and Schmidhuber in 1997 [4]. A good presentation on the principles of meta-learning can be found in Schmidhueber, et al. in 1996 [7] This list is by no means exhaustive, but it should give a starting point for anyone interested in this topic. References: [1] N. E. Cotter and P. R. Conwell. "Fixed-Weight Networks Can Learn." In International Joint Conference on Neural Networks held in San Diego 1990, IEEE, New York, 1990, pp. II.553- 559 [2] N. E. Cotter and P. R. Conwell. "Learning Algorithms and Fixed Dynamics." In International Joint Conference on Neural Networks held in Seattle 1991. by IEEE. New York: IEEE 1991, pp. I.799- 804. [3] L. A. Feldkamp, G. V. Puskorius and P. C. Moore. "Adaptation from Fixed-Weight Dynamic Networks. "Proceedings of International Conference of Neural Networks 96, IEEE-1996 [4] Sepp Hochreiter and Juergen Schmidhuber, "Long Short-Term Memory." Neural Computation. 9(8) pp. 1735-1780, 1997 [5] Sepp Hochreiter, A. Steven Younger and Peter R. Conwell, "Learning to learn using gradient descent", Lecture Notes on Comp. Sci. 2130, Proc. Intl. Conf. on Artificial Neural Networks (ICANN-2001), editors G. Dorffner and H. Bischof and K. Hornik", Springer: Berlin, Heidelberg, pp. 87-94, 2001 [6] Juergen Schmidhuber. "A neural network that embeds its own meta-levels." In Proc. Of the International Conference on Neural Networks '93, San Fransisco, IEEE 1993 [7] Juergen Schmidhuber, Jieyu Zhao, and Marco Wiering. "Simple Principles of Meta-Learning", TR-IDSIA-69-96 http://www.idsia.ch [8] A. Steven Younger, P. R. Conwell, and N. E. Cotter. "Fixed-Weight On-Line Learning." IEEE Transactions on Neural Networks. Vol.10 No. 2, March 1999 pp. 272-283 [9] A. Steven Younger, Sepp Hochreiter, and Peter R. Conwell. "Meta-Learning with Backpropagation." IJCNN'01 International Joint Conference on Neural Networks 2001. IEEE-2001, pp.2001-2006 ------------------------------------------------------------------------- A. Steven Younger syounger at boulder.net Department of Computer Science Data Fusion Corporation University of Colorado at Boulder Northglenn CO Boulder CO From M.Garagnani at open.ac.uk Mon Sep 3 00:38:25 2001 From: M.Garagnani at open.ac.uk (M.Garagnani@open.ac.uk) Date: Mon, 3 Sep 2001 05:38:25 +0100 Subject: Research Fellowship in Spatial Reasoning and AI Planning Message-ID: ( Apologies if you receive multiple copies ) ( Please forward to anyone who could be interested ) ================================================== Computing Department Faculty of Mathematics and Computing The Open University, United Kingdom ---------------------------------------------------- RESEARCH FELLOW IN SPATIAL REASONING AND AI PLANNING The AI (Artificial Intelligence) group of the Computing Department at the Open University is looking for a Research Fellow to work on a new project to investigate the use of spatial reasoning and pattern-matching techniques in AI planning. The project, funded by the Engineering and Physical Sciences Research Council (EPSRC), is due to start in November 2001. The post is for 16 months. The overall aim of the project will be to demonstrate the viability of an entirely new approach to AI planning, based on multi-dimensional spatial representations of domains and problems. If successful, the results of this project are expected to mark the beginning of a new, interdisciplinary, area of research. Candidates should hold a PhD in Computing (or be close to completion) and have experience in the field of Spatial Reasoning. Knowledge of connectionist systems and/or AI planning would be an advantage. The main deliverable of the project will be a new planning system prototype which will make use of pattern-matching and spatial reasoning techniques to represent and reason about actions and states of the world. The successful candidate will take up the post as soon as possible and will be based in the Computing Department of the Faculty of Mathematics and Computing, the Open University, Milton Keynes (United Kingdom). By joining the AI Research Group, a node of the European Network of Excellence in AI Planning (see http://www.planet-noe.org/index.html), the postholder will become part of a dynamic and stimulating research team and will be able to contribute actively to the shaping of this rapidly expanding group. To learn more about the general context of the research, please look at the web site http://computing.open.ac.uk/ or telephone Prof Pat Hall on +44 (0)1825 71 2661 or +44 (0)1908 652694, email p.a.v.hall at open.ac.uk. Appointment will be made on the Research Fellow 1A salary scale in the range 17,451-21,290. For further particulars, an application pack and access details for disabled applicants, please contact the Recruitment Secretary (mcs-recruitment at open.ac.uk), or telephone +44 (0)1908 654161. Please quote ref. 7561. The closing date for applications is Friday 28th September 2001. Interviews will be on 10th October 2001. Disabled applicants whose skills and experience meet the requirements of the job will be interviewed. Please let us know if you need your copy of the further particulars in large print, on computer disk, or on audio cassette tape. Hearing impaired persons may make enquiries on Milton Keynes +44 (0)1908 654901 (Minicom answerphone). The University offers a wide range of jobs with excellent training and career development opportunities. We actively promote equal opportunities in education and employment and welcome applications from all sections of the community. =================== Dr. Max Garagnani Visiting Scholar International Computer Science Institute 1947 Center St. Berkeley, CA 94704-1198 Tel. +1 510 666 2926 Fax. +1 510 666 2956 From harnad at coglit.ecs.soton.ac.uk Mon Sep 3 06:35:46 2001 From: harnad at coglit.ecs.soton.ac.uk (Stevan Harnad) Date: Mon, 3 Sep 2001 11:35:46 +0100 (BST) Subject: Two Visual Systems: BBS Call for Commentators Message-ID: Below is the abstract of a forthcoming BBS target article Please reply to: calls at bbsonline.org Two Visual Systems and Two Theories of Perception: An Attempt to Reconcile the Constructivist and Ecological Approaches by Joel Norman http://www.bbsonline.org/Preprints/Norman/ http://psy.haifa.ac.il/~maga/tvs&ttp.pdf This article has been accepted for publication in Behavioral and Brain Sciences (BBS), an international, interdisciplinary journal providing Open Peer Commentary on important and controversial current research in the biobehavioral and cognitive sciences. Commentators must be BBS Associates or nominated by a BBS Associate. To be considered as a commentator for this article, to suggest other appropriate commentators, or for information about how to become a BBS Associate, please reply by EMAIL within three (3) weeks to: calls at bbsonline.org If you are not a BBS Associate, please approach a current BBS Associate (there are currently over 10,000 worldwide) who is familiar with your work to nominate you. All past BBS authors, referees and commentators are eligible to become BBS Associates. A full electronic list of current BBS Associates is available at this location to help you select a name: http://www.bbsonline.org/Instructions/assoclist.html If no current BBS Associate knows your work, please send us your Curriculum Vitae and BBS will circulate it to appropriate Associates to ask whether they would be prepared to nominate you. (In the meantime, your name, address and email address will be entered into our database as an unaffiliated investigator.) To help us put together a balanced list of commentators, please give some indication of the aspects of the topic on which you would bring your areas of expertise to bear if you were selected as a commentator. To help you decide whether you would be an appropriate commentator for this article, an electronic draft is retrievable from the online BBSPrints Archive, at the URL that follows the abstract below. _____________________________________________________________ Two Visual Systems and Two Theories of Perception: An Attempt to Reconcile the Constructivist and Ecological Approaches Joel Norman Department of Psychology University of Haifa Haifa, Israel jnorman at psy.haifa.ac.il KEYWORDS: Visual perception theories, ecological, constructivist, two visual systems, space perception, size perception, dual-process approach ABSTRACT: The two contrasting theoretical approaches to visual perception, the constructivist and the ecological, are briefly presented and illustrated through their analyses of space perception and size perception. Earlier calls for their reconciliation and unification are reviewed. Neurophysiological, neuropsychological, and psychophysical evidence for the existence of two quite distinct visual systems, the ventral and the dorsal, is presented. These two perceptual systems differ in their functions; the ventral system^Rs central function is that of identification, while the dorsal system is mainly engaged in the visual control of motor behavior. The strong parallels between the ecological approach and the functioning of the dorsal system and between the constructivist approach and the functioning of the ventral system are noted. It is also shown that the experimental paradigms used by the proponents of these two approaches match the functions of the respective visual systems. A dual-process approach to visual perception emerges from this analysis, with the ecological-dorsal process transpiring mainly without conscious awareness, while the constructivist-ventral process is normally conscious. Some implications of this dual-process approach to visual-perceptual phenomena are presented, with emphasis on space perception. http://www.bbsonline.org/Preprints/Norman/ http://psy.haifa.ac.il/~maga/tvs&ttp.pdf ___________________________________________________________ Please do not prepare a commentary yet. Just let us know, after having inspected it, what relevant expertise you feel you would bring to bear on what aspect of the article. We will then let you know whether it was possible to include your name on the final formal list of invitees. _______________________________________________________________________ *** SUPPLEMENTARY ANNOUNCEMENTS *** (1) The authors of scientific articles are not paid money for their refereed research papers; they give them away. What they want is to reach all interested researchers worldwide, so as to maximize the potential research impact of their findings. Subscription/Site-License/Pay-Per-View costs are accordingly access-barriers, and hence impact-barriers for this give-away research literature. There is now a way to free the entire refereed journal literature, for everyone, everywhere, immediately, by mounting interoperable university eprint archives, and self-archiving all refereed research papers in them. Please see: http://www.eprints.org http://www.openarchives.org/ http://www.cogsci.soton.ac.uk/~harnad/Tp/nature4.htm --------------------------------------------------------------------- (2) All authors in the biobehavioral and cognitive sciences are strongly encouraged to self-archive all their papers in their own institution's Eprint Archives or in CogPrints, the Eprint Archive for the biobehavioral and cognitive sciences: http://cogprints.soton.ac.uk/ It is extremely simple to self-archive and will make all of our papers available to all of us everywhere, at no cost to anyone, forever. Authors of BBS papers wishing to archive their already published BBS Target Articles should submit it to BBSPrints Archive. Information about the archiving of BBS' entire backcatalogue will be sent to you in the near future. Meantime please see: http://www.bbsonline.org/help/ and http://www.bbsonline.org/Instructions/ --------------------------------------------------------------------- (3) Call for Book Nominations for BBS Multiple Book Review In the past, Behavioral and Brain Sciences (BBS) had only been able to do 1-2 BBS multiple book treatments per year, because of our limited annual page quota. BBS's new expanded page quota will make it possible for us to increase the number of books we treat per year, so this is an excellent time for BBS Associates and biobehavioral/cognitive scientists in general to nominate books you would like to see accorded BBS multiple book review. (Authors may self-nominate, but books can only be selected on the basis of multiple nominations.) It would be very helpful if you indicated in what way a BBS Multiple Book Review of the book(s) you nominate would be useful to the field (and of course a rich list of potential reviewers would be the best evidence of its potential impact!). *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* From glenn at bioss.sari.ac.uk Mon Sep 3 09:31:46 2001 From: glenn at bioss.sari.ac.uk (Glenn Marion) Date: Mon, 03 Sep 2001 14:31:46 +0100 Subject: JOB: Machine learning (2 posts) - Edinburgh Scotland Message-ID: <3B938642.A08E00C8@bioss.ac.uk> BIOMATHEMATICS AND STATISTICS SCOTLAND (BioSS) Permanent Position based in Edinburgh STATISTICAL AND MACHINE LEARNING TECHNIQUES FOR MOLECULAR DATA We are seeking an established research scientist (at least 5 years postgraduate experience and preferably a PhD) with strong mathematical/computational background and good communication skills. The position is permanent and has a starting salary of ?25000 - ?28800, though a higher salary may be available for an exceptional candidate. A further, fixed-term, position may be available for a new post-doctoral researcher (salary up to ?23700). Information on these and several other posts is available from http://www.bioss.sari.ac.uk/vacancies.htm An application form and further information about BioSS can be obtained from the BioSS Administrative Officer, Betty Heyburn, King's Buildings (JCMB), Edinburgh EH9 3JZ, betty at bioss.ac.uk, tel. +44 (0)131 650 4900. Informal enquiries can be made to Chris Glasbey, email chris at bioss.ac.uk, tel. +44 (0)131 650 4899. CLOSING DATE FOR APPLICATIONS: 24 SEPTEMBER 2001 -- ________________________________________________________________________ Glenn Marion Biomathematics & Statistics Scotland, The University of Edinburgh, James Clerk Maxwell Building, The King's Buildings Edinburgh EH9 3JZ glenn at bioss.ac.uk http://www.bioss.ac.uk/~glenn Tel : +44 (0)131 650 4898 Fax : +44 (0)131 650 4901 ________________________________________________________________________ From dprokhor at ford.com Mon Sep 3 11:12:59 2001 From: dprokhor at ford.com (Prokhorov, Danil (D.V.)) Date: Mon, 3 Sep 2001 11:12:59 -0400 Subject: meta-learning/fixed-weight learning without LSTM? Message-ID: <200109031513.f83FD3w25708@dymwsm09.mailwatch.com> Dear Connectionists, After two very recent messages of Juergen Schmidhuber and A. Steven Younger drawing your attention to meta-learning and fixed-weight learning, I thought I would provide you with yet another reference and the abstract of our recent paper fixed-weight learning published in Proceedings of the 11th Yale Workshop on Adaptive and Learning Systems (June 4-6, 2001). L. A. Feldkamp, D. V. Prokhorov, and T. M. Feldkamp, Conditioned Adaptive Behavior from a Fixed Neural Network, Proceedings of the Eleventh Yale Workshop on Adaptive and Learning Systems, New Haven, CT, pp. 78-83, 2001. Abstract (sorry for its conciseness) We demonstrate that a fixed-weight recurrent neural network (RMLP-style) can be trained to exhibit input-output behavior that depends on which of two conditioning tasks had been performed a substantial number of time steps in the past. This behavior can also be made to survive an intervening interference task. ____________________________________________________________________________ It may be difficult to find these Proceedings though. Please contact Danil Prokhorov (dprokhor at ford.com) or Lee Feldkamp (lfeldkam at ford.com) to receive an electronic copy of this paper. Sincerely, Danil Prokhorov Artificial Neural Systems and Fuzzy Logic Group Ford Research Laboratory 2101 Village Rd., MD 2036 Dearborn, MI 48121-2053, U.S.A. From S.M.Bohte at cwi.nl Tue Sep 4 09:12:44 2001 From: S.M.Bohte at cwi.nl (Sander Bohte) Date: Tue, 4 Sep 2001 15:12:44 +0200 Subject: preprint: paper on error-backpropagation in spiking neural networks Message-ID: Apologies if you receive this posting more than once. A preprint is available for download, of the paper 'SpikeProp: Error-Backpropagation for Networks of Spiking Neurons', by Sander Bohte, Joost Kok and Han La Poutr? (to appear in `Neurocomputing'). For a network of spiking neurons that encodes information in the timing of individual spike-times, we derive a supervised learning rule, SpikeProp, akin to traditional error-backpropagation and show how to overcome the discontinuities introduced by thresholding. Using this learning algorithm, we demonstrate how networks of spiking neurons with biologically reasonable action potentials can perform complex non-linear classification in fast temporal coding just as well as rate-coded networks. We perform experiments for the classical XOR-problem, when posed in a temporal setting, as well as for a number of other benchmark datasets. When comparing the (implicit) number of biological neurons that would be required for the respective encodings, it is empirically demonstrated that temporal coding potentially requires significantly less neurons. As we show that reliable temporal computation can only be accomplished by spike-response functions with a time constant longer than the coding interval, the results also refute the long-standing argument against temporal coding that states that the typical integration times of real neurons are too long to compute a fast temporal code. keywords: spiking neural networks, temporal coding, error-backpropagation, XOR. Download Instructions: Go to http://www.cwi.nl/~sbohte/pub_spikeprop.htm and click on the file to download (in PDF format [393K], or zipped Postscript [515K]). Comments Welcome If you have problems downloading, please e-mail me. Sander Bohte From samengo at cab.cnea.gov.ar Tue Sep 4 15:16:28 2001 From: samengo at cab.cnea.gov.ar (Ines Samengo) Date: Tue, 04 Sep 2001 16:16:28 -0300 Subject: two papers: Information maps, and Information loss in decoding Message-ID: <3B95288C.B221002D@cab.cnea.gov.ar> Dear connectionists, The following papers may be of interest to you: Measuring information spatial densities Michele Bezzi, Ines Samengo, Stefan Leutgeb and Sheri J. Mizumori Accepted in Neural Computation, 2001. Abstract: A novel definition of the stimulus-specific information is presented, which is particularly useful when the stimuli constitute a continuous and metric set, as for example, position in space. The approach allows one to build the spatial information distribution of a given neural response. The method is applied to the investigation of putative differences in the coding of position in hippocampus and lateral septum. http://www.cab.cnea.gov.ar/users/samengo/pub.html ---------------- The information loss in an optimal maximum likelihood decoding Ines Samengo Accepted in Neural Computation, 2001. Abstract: The mutual information between a set of stimuli and the elicited neural responses is compared to the corresponding decoded information. The decoding procedure is presented as an artificial flattening of the joint probabilities between stimuli and responses. The information loss is quantified. Whenever the conditional probabilities are only slightly distorted, the information loss is shown to be quadratic in the distortion. http://www.cab.cnea.gov.ar/users/samengo/pub.html _______________________________________________________________ Ines Samengo samengo at cab.cnea.gov.ar fax: +54 2944 445299 Centro Atomico Bariloche (8.400) San Carlos de Bariloche Rio Negro Argentina _______________________________________________________________ From matsudaira at gmd.gr.jp Tue Sep 4 22:47:06 2001 From: matsudaira at gmd.gr.jp (Yukako Matsudaira) Date: Wed, 5 Sep 2001 11:47:06 +0900 Subject: Researchers Post Offerings for GMD Japan Research Laboratory Message-ID: <006501c135b5$0d2d2660$0202630a@yukakopc> Research Positions - Announcement ---------------------------------------------------------------------- Research Positions in Autonomous Robotics in Open and Dynamic Environments ------------------------------------------------------------------------ Seven new research positions (including post-doctoral positions) are open at GMD-Japan Research Laboratory, Kitakyushu, Japan and will be filled at the earliest convenience. The laboratory (planned to be a team of eight scientists and two support staff) is based on long-term cooperations with the Japanese research community and the (local) companies and organizations. It focuses on robotics in two main research fields, namely autonomous inspection of sewer pipes and soccer-playing robot teams, as known from the RoboCup competitions. Investigated questions are: - How to self-localize and move in many DoF without global correlation? - Interpretation / integration / abstraction / compression of complex sensor signals? - How to build models based upon sensor-motor coordination for (partially) unknown environments? - How to predict sensor-actuator correlations and generate motor signals appropriately? - How to coordinate multiple loosely coupled robots? Autonomous mobile robotics for sewer inspection and soccer-playing are regarded as one of the most promising experimental and application environment in this context. Real dynamic environments and real autonomy (which is required in these setups), settle these questions in concrete environments. The overall goal is of course not only implementing prototype systems, but to get a better understanding of autonomy, and situatedness. Modeling, adaptation, clustering, prediction, communication, or - from the perspective of robotics - spatial and behavioral modeling, localization, navigation, and exploration are cross-topics addressed in most questions. It is also very important to connect this with possible applications to be explored with (local) companies, like sewer operational units or edutainment robotics manufacturers. Techniques to be employed and developed include dynamical systems, connectionist approaches, behavior-based techniques, rule based systems, and systems theory. Experiments are based on physical robots. Thus the discussion of experimental setups and particularly the meaning of embodiment and situatedness are natural topics. GMD-JRL is an independent research unit embedded and located at the newly opened Science and Research Park (run by the Kitakyushu Foundation for the Advancement of Industry, Science and Technology FAIS http://www.city.kitakyushu.jp). Kitakyushu is located at the coast line of the Sea of Japan on Kyushu Island. It is as far away from Tokyo as from Shanghai and a traditional harbour gate to south-east Asia. In addition, there are close connections to the Fraunhofer Institute of Autonomous Intelligent Systems (FhI AIS, http://www.fraunhofer.ais.gmd.de/ ) headed by Prof. Dr. Thomas Christaller in Sankt Augustin, Germany. The technical environment is up-to-date and the laboratory itself is located in four modern office rooms on the campus of the Science and Research Park, which can be reached by a 30 minutes ride from central Kitakyushu. If the above challenges arise your interest, please proceed to our expectations regarding the ideal candidate: - Master/Diploma or Ph.D. / doctoral degree in computer sciences, electrical or mechanical engineering, physics, mathematics, biology, or related disciplines: - Experiences in experimenting with autonomous systems. - Theoretical foundations in mathematics, control theory, connectionism, dynamical systems, or systems theory. - Interest in joining an international team of highly motivated researchers. Furthermore it is expected that the candidate engages her/himself to bring in her/his own perspective to blend it with the whole group. There are many opportunities for writing a Ph.D. thesis. Salary starts at 7.8 Mill. Yen per year depending on experience and graduation. For any further information, and applications (including addresses of referees, two recent publications, and a letter of interest) please contact: Prof. Dr. Thomas Christaller, Director of GMD-Japan Research Laboratory Secretary: Yukako Matsudaira mailto:matsudaira at gmd.gr.jp Tel: +81-93-695-3085 Fax: +81-93-695-3525 URL: http://www.gmd.de/JRL/ -- _______________________________________________________ Visit the AiS Colloquium on Autonomy every two weeks on Wednesday: http://www.ais.fraunhofer.de/kolloquium/index.html Visit the European Conference on Artificial Life at Prague http://www.cs.cas.cz/ecal2001 Visit the web portal on Edutainment Robotics http://www.edutainment-robotics.org Visit the RoboFesta in Japan http://www.robofesta.net Notice the Interdisciplinary College at G=FCnne http://www.tzi.de/ik2002 _______________________________________________________ Prof. Dr. rer.nat. Thomas Christaller mailto:thomas.christaller at ais.fraunhofer.de http://www.ais.fraunhofer.de/people/Thomas.Christaller/ Fraunhofer Institute for Autonomous intelligent Systems - AiS http://www.ais.fraunhofer.de/ Schloss Birlinghoven D-53754 Sankt Augustin Phone: +49-2241-14-2678, Fax +49-2241-14-2384 Secretary: mailto:Renate.Henkeler at ais.fraunhofer.de, mailto:Eva.Sommer at ais.fraunhofer.de _______________________________________________________ From aapo at james.hut.fi Wed Sep 5 06:37:38 2001 From: aapo at james.hut.fi (Aapo Hyvarinen) Date: Wed, 5 Sep 2001 13:37:38 +0300 Subject: New Book on ICA Message-ID: [Our apologies if you receive multiple copies of this message.] The following book is now available: INDEPENDENT COMPONENT ANALYSIS Aapo Hyvarinen, Juha Karhunen, and Erkki Oja Published by John Wiley & Sons, 2001. 504 pages. Independent Component Analysis (ICA) is one of the most exciting topics in the fields of neural computation, advanced statistics, and signal processing. This is the first book to provide a comprehensive introduction to this new technique complete with the mathematical background needed to understand and utilize it. It offers a general overview of the basics of ICA, important solutions and algorithms, and in-depth coverage of new applications in vision research, brain imaging, telecommunications, and more. See the book web page at http://www.cis.hut.fi/projects/ica/book/ ---------------------------------------------------- Aapo Hyvarinen Neural Networks Research Centre Helsinki University of Technology P.O.Box 5400, FIN-02015 HUT, Finland Tel: +358-9-4513278, Fax: +358-9-4513277 Email: Aapo.Hyvarinen at hut.fi Home page: http://www.cis.hut.fi/aapo/ ---------------------------------------------------- From jianfeng at cogs.susx.ac.uk Wed Sep 5 09:53:32 2001 From: jianfeng at cogs.susx.ac.uk (Jianfeng Feng) Date: Wed, 5 Sep 2001 14:53:32 +0100 (BST) Subject: Ph.D studentship Message-ID: We are interested in applicants having backgrounds in any of the following Statistics Mathematics Computer Science Physics Sussex has an excellent research reputation (12th in the latest league table). The school of cognitive and computing sciences at the university of Sussex grew out of a pioneering multi-disciplinary centre for research into intelligent systems and the mechanisms underlying them. Artificial intelligence and Neuroscience are areas in which Sussex is exceptionally strong, a view supported by the results of the last three national Research Assessment Exercises in which the computer science and artificial intelligence group in the school received 5 rating. Applications are invited for one research studentship to be held from 1st January 2002 for three years. The studentship is funded by EPSRC. The student will work on learning in spiking neuronal networks. Applicants should have a 1st or 2(i) class honours degree in an appropriate subject. Further particulars on my interests etc. could be found at http://www.cogs.susx.ac.uk/users/jianfeng Applications should be sent as soon as possible to: Dr. J. Feng, Cogs, Sussex University, BN1 9QH, UK, e-mail: jianfeng at cogs.susx.ac.uk, or jf218 at cam.ac.uk, tel. 0044 1273 678062. From smyth at ics.uci.edu Thu Sep 6 13:35:22 2001 From: smyth at ics.uci.edu (Padhraic Smyth) Date: Thu, 06 Sep 2001 10:35:22 -0700 Subject: faculty positions in new Statistics Dept at UCI Message-ID: <3B97B3DA.CBDD94B0@ics.uci.edu> [apologies as usual if you receive this more than once] Dear Colleagues, The University of California, Irvine (UCI) has just announced the creation of a new statistics department. In the coming year UCI will be hiring for three positions, two tenured (one of which is the chair) and one tenure-track. Computational statistics (in a broad sense, including graphical models, machine learning, etc) is one of the areas where we are likely to be hiring, so I would encourage those of you who are interested to please apply and/or to pass the word along to any colleagues who might be interested. (full details below). This new department is likely to offer significant opportunities for interdisciplinary collaborations with faculty and students in other departments at UCI including computer science, biological sciences, medicine, cognitive science, economics, all of which are actively involved in setting up the new department. Padhraic Smyth Information and Computer Science University of California, Irvine smyth at ics.uci.edu UNIVERSITY OF CALIFORNIA, IRVINE Announces a Department in Statistics The University of California, Irvine is starting a Department of Statisitcs. We anticipate appointing a full-time faculty in statistics of 6-8 people over the next seven years, with several more half-time appointments shared with other units at UCI. It will be a department with a strongly interdisciplinary flavor, focused both on theoretical research and applied problems, and it will be associated with an independent statistical consulting center that presently serves the campus and surrounding businesses and industry. Three faculty positions are open for recruitment in 2001-02: two with tenure, including one for the Chair of the Department; and one tenure-track assistant professorship. In all cases demonstrated excellence in research, teaching, and service are sought and, in the case of the Chair evidence of leadership ability is required. UCI is one of the youngest campuses in the University of California, yet we are already ranked 10th among public universities. We are projected to grow by almost 50% over the next ten years, with significant increases in graduate enrollment. The new Department of Statistics will occupy a prominent place in this expanding academic profile, and it will interact significantly with a wide range of existing departments and innovative programs across the whole campus. A detailed account of plans for the new department is available at www.evc.uci.edu/proposal.pdf. For information about UCI, see www.uci.edu/. For information about the community around UCI, see www.oc.ca.gov/. Completed applications with a cover letter, sample research publications, and if possible three letters of recommendation and up to five additional names who may be contacted should be sent to the Chair of the search committee: Professor Duncan Luce c/o Office of the Executive Vice Chancellor 535 Administration Bldg. University of California, Irvine Irvine, CA 92697-1000 (Applications received before November 1, 2001, will receive preference, but later applications will be considered.) The University of California is an Equal Opportunity Employer committed to excellence through diversity. From P.J.Lisboa at livjm.ac.uk Fri Sep 7 04:45:28 2001 From: P.J.Lisboa at livjm.ac.uk (Lisboa Paulo) Date: Fri, 7 Sep 2001 09:45:28 +0100 Subject: Workshop on Regulatory Issues in Medical Decision Support, UCL, F riday 19th October 2001 Message-ID: Dear Colleagues, Europe and the USA have rapidly growing markets for medical equipment, whose operation increasingly relies on complex software. The explosion of networking and databases, coupled with user demands for better quality, personalisation and remote care, all point to a fast commercial development of healthcare informatics. However, there are many unresolved issues surrounding the application of the Medical Devices Directives, especially in Medical Decision Support. This workshop brings together the main complementary perspectives that impact directly on regulatory procedures, introduced by speakers who are recognised authorities in the design, evaluation and practical use of statistical, rule-based and neural network systems. The purpose of the workshop is to identify and discuss key issues, so as to provide pointers to effective and practical ways to resolve them. A programme and registration leaflet is attached. In time a link will appear also at the AIME website http://www.aime.org.uk/ Paulo Lisboa. ____________________________________________________________________________ PROGRAMME 09.30-10.15 Registration and Coffee 10.25-10.30 Welcome Professor Paulo Lisboa, John Moores University, Liverpool. 10.30-11.10 The Medical Devices Directives: Application of the Directives to Medical Decision Support Systems (Scope and Conformity Assessment) Mr. P. Stonebrook, Medical Devices Agency. 11.10-11.50 Key issues for computer-based decision support in clinical practice Dr. Bipin Vadher, Bromley Hospitals NHS Trust. 11.50-12.30 Experience with longstanding decision support and perspectives of new developments in Europe. Dr. Susan Clamp, Clinical Information Science Unit, Leeds. 12.30-13.30 Lunch 13.30-14.15 Issues in the development of good statistical models for prognosis Professor Doug Altman, Centre for Statistics in Medicine, Oxford. 14.15-15.00 Clinical Decision- making by machine: maximising safety and limiting risk Professor John Fox, ICRF, Lincoln's Inn Fields. 15.00-15.15 Tea 15.15-16.00 Validation of neural network medical systems Dr. Ian Nabney, Cardionetics Institute of Bioinformatics, Aston University. 16.00-16.30 Discussion Dr. Jeremy Wyatt, UCL, to lead on practical ways forward in the evaluation and certification of software for medical decision support. 16.30 Close of Meeting ____________________________________________________________________________ _______ REGISTRATION LEAFLET Regulatory Issues in Medical Decision Support 19th October, University College London REGISTRATION FEES:(Incl. of lunch and coffee) Members of AIME Institutions and BMIS ? 60.00 Non-Members ? 75.00 Student / Retired ? 40.00 Special requirements (e.g. dietary, disabled) Title: Name: Address: Post Code: Tel: Fax: Email: METHOD OF PAYMENT (Cheques payable to 'The Institute of Physics & Engineering in Medicine') I wish to attend the above meeting and enclose my registration fee. I wish to attend the above meeting. Please debit my credit card as detailed below. (Barclaycard / Access / Visa / Mastercard / Eurocard) Expiry Date / Signature I wish to attend the above meeting. Please send an invoice for my registration fee to the address given below. My Purchase Order Number is _________________ Please note that invoices cannot be supplied without an official purchase order. PLEASE NOTE THAT REGISTRATIONS CANNOT BE ACCEPTED WITHOUT EITHER ADVANCE PAYMENT OR AN OFFICIAL PURCHASE ORDER. Please return completed form to:IPEM Meetings Fairmount House, 230 Tadcaster Road, York YO24 1ES Fax: 01904 612279 To be received no later than 5 October 2001. From kstanley at cs.utexas.edu Fri Sep 7 17:08:06 2001 From: kstanley at cs.utexas.edu (Kenneth Owen Stanley) Date: Fri, 7 Sep 2001 16:08:06 -0500 (CDT) Subject: Neuroevolution paper, software, and demo Message-ID: <200109072108.QAA22092@wheat.cs.utexas.edu> Dear Connectionists, NeuroEvolution of Augmenting Topologies (NEAT) is a new approach to evolving the topology and weights of artificial neural networks. Neural network topologies in NEAT start minimally and grow increasingly complex over generations. A paper describing the method (abstract below), source code, and animated demos in a robot control domain are all available at: http://www.cs.utexas.edu/users/nn/pages/research/ne-methods.html#NEAT --Kenneth Stanley and Risto Miikkulainen Paper: ----------------------------------------------------------------------- EVOLVING NEURAL NETWORKS THROUGH AUGMENTING TOPOLOGIES Kenneth O. Stanley and Risto Miikkulainen Department of Computer Sciences, The University of Texas at Austin Technical Report TR-AI-01-290, June 2001. http://www.cs.utexas.edu/users/nn/pages/publications/abstracts.html#stanley.utcstr01.ps.gz An important question in neuroevolution is how to gain an advantage from evolving neural network topologies along with weights. We present a method, NeuroEvolution of Augmenting Topologies (NEAT) that outperforms the best fixed-topology method on a challenging benchmark reinforcement learning task. We claim that the increased efficiency is due to (1) employing a principled method of crossover of different topologies, (2) protecting structural innovation using speciation, and (3) incrementally growing from minimal structure. We test this claim through a series of ablation studies that demonstrate that each component is necessary to the system as a whole and to each other. What results is significantly faster learning. NEAT is also an important contribution to GAs because it shows how it is possible for evolution to both optimize and complexify solutions simultaneously, offering the possibility of evolving increasingly complex solutions over generations, and strengthening the analogy with biological evolution. Software: ----------------------------------------------------------------------- NEAT C++ SOURCE CODE http://www.cs.utexas.edu/users/nn/pages/software/abstracts.html#neat-cpp Kenneth Stanley The NEAT package contains source code implementing the NeuroEvolution of Augmenting Topologies method. The source code is written in C++. The package includes implementations of experiments for XOR, single pole balancing, and both Markovian and non-Markovian double pole balancing. A 17-page postscript documentation file is included to make getting started easier. Demo: ----------------------------------------------------------------------- COMPETITIVE COEVOLUTION ROBOT CONTROL DEMO http://www.cs.utexas.edu/users/nn/pages/research/neatdemo.html Kenneth Stanley The page contains links to movies (in the form of GIF animations) depicting evolved robot controllers controlling simulated Khepera-like robots in a real-time pursuit and evasion task. The controllers were evolved using the NEAT method. The movies show the robots' behavior in the task, and also the actual evolved neural network controllers with neurons firing in real-time, so you can see what the robots are "thinking" as they compete. The main point of the demo is to demonstrate that NEAT establishes an evolutionary "arms race", i.e. is able to discover increasingly complex behaviors in competitive coevolution. From Steve.Minton at fetch.com Sun Sep 9 17:20:44 2001 From: Steve.Minton at fetch.com (Steve Minton) Date: Sun, 9 Sep 2001 14:20:44 -0700 Subject: New JAIR paper Message-ID: It was suggested to me that members of this mailing list may be interested in the following announcement: Bhattacharyya, C. and Keerthi, S.S. (2001) "Mean Field Methods for a Special Class of Belief Networks", Volume 15, pages 91-114. Available in PDF, PostScript and compressed PostScript. For quick access via your WWW browser, use this URL: http://www.jair.org/abstracts/bhattacharyya01a.html More detailed instructions are below. Abstract: The chief aim of this paper is to propose mean-field approximations for a broad class of Belief networks, of which sigmoid and noisy-or networks can be seen as special cases. The approximations are based on a powerful mean-field theory suggested by Plefka. We show that Saul, Jaakkola and Jordan' s approach is the first order approximation in Plefka's approach, via a variational derivation. The application of Plefka's theory to belief networks is not computationally tractable. To tackle this problem we propose new approximations based on Taylor series. Small scale experiments show that the proposed schemes are attractive. The article is available via: -- comp.ai.jair.papers (also see comp.ai.jair.announce) -- World Wide Web: The URL for our World Wide Web server is http://www.jair.org/ For direct access to this article and related files try: http://www.jair.org/abstracts/bhattacharyya01a.html -- Anonymous FTP from either of the two sites below. Carnegie-Mellon University (USA): ftp://ftp.cs.cmu.edu/project/jair/volume15/bhattacharyya01a.ps The University of Genoa (Italy): ftp://ftp.mrg.dist.unige.it/pub/jair/pub/volume15/bhattacharyya01a.ps The compressed PostScript file is named bhattacharyya01a.ps.Z (159K) For more information about JAIR, visit our WWW or FTP sites, or contact jair-ed at isi.edu From rampon at tin.it Sun Sep 9 18:02:29 2001 From: rampon at tin.it (Salvatore Rampone) Date: Mon, 10 Sep 2001 00:02:29 +0200 Subject: New genetic database and paper available Message-ID: Dear Connectionists, This new genetic dataset may be of interest to you: Database Name: HS3D - Homo Sapiens Splice Site Dataset URL: http://www.sci.unisannio.it/docenti/rampone/ The data base is described in the paper HS3D - Homo Sapiens Splice Site Dataset by Pollastro, P., Rampone, S. Universit del Sannio - ITALY Accepted in Nucleic Acids Research 2002 Database Issue The extended abstract is available at http://space.tin.it/scienza/srampone/ (publication page) or directly from http://space.tin.it/scienza/srampone/ramp0201.pdf. ------- Abstract: In the last years many computational tools for gene identification and characterization, mostly based on machine learning approaches, have been used. In the machine learning approach, a learning algorithm receives a set of training examples, each labelled as belonging to a particular class. The algorithm's goal is to produce a classification rule for correctly assigning new examples to these classes. The success of these methods depends largely on the quality of the data sets that are used as the training set. Furthermore a common data set is necessary when the prediction accuracy of different programs needs to be comparatively assessed. The Irvine Primate Splice Junctions Dataset (UCI Machine Learning Repository http://www.ics.uci.edu/~mlearn/MLRepository.html) is a standard "de facto" in the machine learning community, but it is now very out of date and does not include sufficient material for the most learning algorithm needs. A recent and EST confirmed data set has the same limitation in the data extend. More recently Burset et al. developed an extensive data base, but the data do not include false splice sites (negative examples), and, specifically, proximal false splice sites. The latter data form a well known critical point of classification systems. We developed a new database (HS3D - Homo Sapiens Splice Site Dataset) of Homo Sapiens Exon, Intron and Splice regions. The aim of this data set is to give standardized material to train and to assess the prediction accuracy of computational approaches for gene identification and characterization. From pam_reinagel at hms.harvard.edu Mon Sep 10 11:49:12 2001 From: pam_reinagel at hms.harvard.edu (Pamela Reinagel) Date: Mon, 10 Sep 2001 11:49:12 -0400 Subject: conference: natural stimulus statistics Message-ID: We are pleased to announce a new Gordon Research Conference Sensory coding and the natural environment: Probabilistic models of perception June 30 - July 5, 2002 Mount Holyoke College, MA. This conference will bring together participants from many disciplines to discuss the statistical structure of natural sensory stimuli, and how biological systems may use these statistics to process natural signals. The meeting grew out of a smaller workshop and will cover similar topics: From 1997: www.klab.caltech.edu/~pam/nssmeeting From 2000: www.klab.caltech.edu/~pam/nss2000.html As the date approaches, details and applications will be available on the GRC website (http://www.grc.uri.edu). Until then, please mark the date on your calendars! Pam Reinagel and Bruno Olshausen co-chairs, 2002 GRC From rampon at tin.it Mon Sep 10 14:01:47 2001 From: rampon at tin.it (Salvatore Rampone) Date: Mon, 10 Sep 2001 20:01:47 +0200 Subject: Paper address correction - New genetic database and paper available Message-ID: Dear Connectionists, with reference to the message "New genetic database and paper available" posted to this list, the paper downlod address is http://space.tin.it/scienza/srampone/ramp0201.zip and not http://space.tin.it/scienza/srampone/ramp0201.pdf. Sorry for the mistake. -salvatore ----------- Salvatore Rampone Facolta' di Scienze MM.FF.NN. and INFM Universita' del Sannio Via Port'Arsa 11 I-82100 Benevento ITALY E-mail: rampone at unisannio.it From: esann To: "Connectionists at cs.cmu.edu" References: From bogus@does.not.exist.com Thu Sep 13 10:49:38 2001 From: bogus@does.not.exist.com () Date: Thu, 13 Sep 2001 16:49:38 +0200 Subject: CFP: ESANN'2002 European Symposium on Artificial Neural Networks Message-ID: ---------------------------------------------------- | | | ESANN'2002 | | | | 10th European Symposium | | on Artificial Neural Networks | | | | Bruges (Belgium) - April 24-25-26, 2002 | | | | Announcement and call for papers | ---------------------------------------------------- Technically co-sponsored by the IEEE Region 8*, the IEEE Benelux Section, the International Neural Networks Society and the European Neural Networks Society (* to be confirmed). The call for papers for the ESANN'2002 conference is now available on the Web: http://www.dice.ucl.ac.be/esann For those of you who maintain WWW pages including lists of related ANN sites: we would appreciate if you could add the above URL to your list; thank you very much! We try as much as possible to avoid multiple sendings of this call for papers; however please apologize if you receive this e-mail twice, despite our precautions. You will find below a short version of this call for papers, without the instructions to authors (available on the Web). If you have difficulties to connect to the Web please send an e-mail to esann at dice.ucl.ac.be and we will send you a full version of the call for papers. ESANN'2002 is organised in collaboration with the UCL (Universite catholique de Louvain, Louvain-la-Neuve) and the KULeuven (Katholiek Universiteit Leuven). Scope and topics ---------------- Since its first happening in 1993, the European Symposium on Artificial Neural Networks has become the reference for researchers on fundamentals and theoretical aspects of artificial neural networks. Each year, around 100 specialists attend ESANN, in order to present their latest results and comprehensive surveys, and to discuss the future developments in this field. The ESANN'2002 conference will focus on fundamental aspects of ANNs: theory, models, learning algorithms, mathematical and statistical aspects, in the context of function approximation, classification, control, time-series prediction, signal processing, vision, etc. Papers on links and comparisons between ANNs and other domains of research (such as statistics, data analysis, signal processing, biology, psychology, evolutive learning, bio-inspired systems, etc.) are encouraged. Papers will be presented orally (no parallel sessions) and in poster sessions; all posters will be complemented by a short oral presentation during a plenary session. It is important to mention that it is the topics of the paper which will decide if it better fits into an oral or a poster session, not its quality. The selection of posters will be identical to oral presentations, and both will be printed in the same way in the proceedings. Nevertheless, authors have the choice to indicate on the author submission form that they only accept to present their paper orally. The following is a non-exhaustive list of topics covered during the ESANN conferences: - Models and architectures - Learning algorithms - Theory - Mathematics - Statistical data analysis - Classification - Approximation of functions - Time series forecasting - Nonlinear dimension reduction - Multi-layer Perceptrons - RBF networks - Self-organizing maps - Vector quantization - Support Vector Machines - Recurrent networks - Fuzzy neural nets - Hybrid networks - Bayesian neural nets - Cellular neural networks - Signal processing - Independent component analysis - Natural and artificial vision - Adaptive control - Identification of non-linear dynamical systems - Biologically plausible networks - Bio-inspired systems - Cognitive psychology - Evolutiv learning - Adaptive behaviour Special sessions ---------------- Special sessions will be organized by renowned scientists in their respective fields. Papers submitted to these sessions are reviewed according to the same rules as any other submission. Authors who submit papers to one of these sessions are invited to mention it on the author submission form; nevertheless, submissions to the special sessions must follow the same format, instructions and deadlines as any other submission, and must be sent to the same address. Here is the list of special sessions that will be organized during the ESANN'2002 conference: 1. Perspectives on Learning with Recurrent Networks (B. Hammer, J.J. Steil) 2. Representation of high-dimensional data (A. Gurin-Dugu, J. Hrault) 3. Neural Network Techniques in Fault Detection and Isolation (S. Simani) 4. Hardware and Parallel Computer Implementations of Neural Networks (U. Seiffert) 5. Exploratory Data Analysis in Medicine and Bioinformatics (A. Wismller, T. Villmann) 6. Neural Networks and Cognitive Science (H. Paugam-Moisy, D. Puzenat) A short description of these sessions will be inserted on the ESANN Web site in the next few days, and tentatively sent to this distribution list. Location -------- The conference will be held in Bruges (also called "Venice of the North"), one of the most beautiful medieval towns in Europe. Bruges can be reached by train from Brussels in less than one hour (frequent trains). The town of Bruges is world-wide known, and famous for its architectural style, its canals, and its pleasant atmosphere. The conference will be organised in a hotel located near the centre (walking distance) of the town. There is no obligation for the participants to stay in this hotel. Hotels of all level of comfort and price are available in Bruges; there is a possibility to book a room in the hotel of the conference at a preferential rate through the conference secretariat. A list of other smaller hotels is also available. The conference will be held at the Novotel hotel, Katelijnestraat 65B, 8000 Brugge, Belgium. Call for contributions ---------------------- Prospective authors are invited to submit six copies of their manuscript (including at least two originals or very good copies without glued material, which will be used for the proceedings) one signed copy of the author submission form a floppy disk or a CD (PC format preferred) containing their contribution in (generic) PostScript format before December 3, 2001. Sorry, electronic or fax submissions are not accepted. The working language of the conference (including proceedings) is English. The instructions to authors, together with the author submission form, are available on the ESANN Web server: http://www.dice.ucl.ac.be/esann Authors must indicate their choice for oral or poster presentation on the author submission form. They must also sign a written agreement that they will register to the conference and present the paper in case of acceptation of their submission. Authors of accepted papers will have to register before February 28, 2002. They will benefit from the advance registration fee. Submissions must be sent to: Michel Verleysen UCL - DICE 3, place du Levant B-1348 Louvain-la-Neuve Belgium esann at dice.ucl.ac.be All submissions will be acknowledged by fax or email before the end of December, 2001. Deadlines --------- Submission of papers December 3, 2001 Notification of acceptance February 3, 2002 Symposium April 24-26, 2002 Registration fees ----------------- registration before registration after March 15, 2002 March 15, 2002 Universities 420 450 Industries 520 550 The registration fee includes the attendance to all sessions, the ESANN'2002 dinner, a copy of the proceedings, daily lunches (24-26 April 2002), and the coffee breaks. Conference secretariat ---------------------- ESANN'2002 d-side conference services phone: + 32 2 730 06 11 24 av. L. Mommaerts Fax: + 32 2 730 06 00 B - 1140 Evere (Belgium) E-mail: esann at dice.ucl.ac.be http://www.dice.ucl.ac.be/esann Steering and local committee (to be confirmed) ---------------------------- Hugues Bersini Univ. Libre Bruxelles (B) Franois Blayo Prfigure (F) Marie Cottrell Univ. Paris I (F) Jeanny Hrault INPG Grenoble (F) Bernard Manderick Vrije Univ. Brussel (B) Eric Noldus Univ. Gent (B) Jean-Pierre Peters FUNDP Namur (B) Joos Vandewalle KUL Leuven (B) Michel Verleysen UCL Louvain-la-Neuve (B) Scientific committee (to be confirmed) -------------------- Edoardo Amaldi Politecnico di Milano (I) Herv Bourlard IDIAP Martigny (CH) Joan Cabestany Univ. Polit. de Catalunya (E) Colin Campbell Bristol Univ. (UK) Stphane Canu Inst. Nat. Sciences App. (F) Holk Cruse Universitt Bielefeld (D) Eric de Bodt Univ. Lille II & UCL Louv.-la-N. (B) Dante Del Corso Politecnico di Torino (I) Wlodek Duch Nicholas Copernicus Univ. (PL) Marc Duranton Philips / LEP (F) Richard Duro Univ. Coruna (E) Jean-Claude Fort Universit Nancy I (F) Colin Fyfe Univ. Paisley (UK) Stan Gielen Univ. of Nijmegen (NL) Marco Gori Univ. Siena (I) Bernard Gosselin Fac. Polytech. Mons (B) Manuel Grana UPV San Sebastian (E) Anne Gurin-Dugu INPG Grenoble (F) Barbare Hammer Univ. of Osnabruck (D) Martin Hasler EPFL Lausanne (CH) Laurent Hrault CEA-LETI Grenoble (F) Gonzalo Joya Univ. Malaga (E) Christian Jutten INPG Grenoble (F) Juha Karhunen Helsinky Univ. of Technology (FIN) Vera Kurkova Acad. of Science of the Czech Rep. (CZ) Jouko Lampinen Helsinki Univ. of Tech. (FIN) Petr Lansky Acad. of Science of the Czech Rep. (CZ) Mia Loccufier Univ. Gent (B) Erzsebet Merenyi Rice Univ. (USA) Jean Arcady Meyer Univ. Pierre et Marie Curie - Paris 6 (F) Jos Mira UNED (E) Jean-Pierre Nadal Ecole Normale Suprieure Paris (F) Gilles Pags Univ. Pierre et Marie Curie - Paris 6 (F) Thomas Parisini Politecnico di Milano (I) Hlne Paugam-Moisy Univ. Lumire Lyon 2 (F) Alberto Prieto Universitad de Granada (E) Leonardo Reyneri Politecnico di Torino (I) Tamas Roska Hungarian Academy of Science (H) Jean-Pierre Rospars INRA Versailles (F) Jose Santos Reyes Univ. Coruna (E) Jochen Steil Univ. Bielefeld (D) John Stonham Brunel University (UK) Johan Suykens KUL Leuven (B) John Taylor Kings College London (UK) Claude Touzet IUSPIM Marseilles (F) Marc Van Hulle KUL Leuven (B) Thomas Villmann Univ. Leipzig (D) Christian Wellekens Eurecom Sophia-Antipolis (F) ======================================================== ESANN - European Symposium on Artificial Neural Networks http://www.dice.ucl.ac.be/esann * For submissions of papers, reviews,... Michel Verleysen Univ. Cath. de Louvain - Microelectronics Laboratory 3, pl. du Levant - B-1348 Louvain-la-Neuve - Belgium tel: +32 10 47 25 51 - fax: + 32 10 47 25 98 mailto:esann at dice.ucl.ac.be * Conference secretariat d-side conference services 24 av. L. Mommaerts - B-1140 Evere - Belgium tel: + 32 2 730 06 11 - fax: + 32 2 730 06 00 mailto:esann at dice.ucl.ac.be ======================================================== From felix at idsia.ch Fri Sep 14 10:40:09 2001 From: felix at idsia.ch (Felix Gers) Date: Fri, 14 Sep 2001 16:40:09 +0200 (CEST) Subject: LSTM recurrent nets, PhD Thesis, Papers & Code Message-ID: Dear Connectionists, I am glad to announce my PhD thesis on Long Short-Term Memory (LSTM) in Recurrent Neural Networks (RNNs), several LSTM papers, and LSTM source code. Felix Gers, IDSIA www.idsia.ch -------------------------------PHD THESIS------------------------------ Long Short-Term Memory in Recurrent Neural Networks: http://www.idsia.ch/~felix/My_papers/phd.ps.gz http://www.idsia.ch/~felix/My_papers/phd.pdf On-line abstract: http://www.idsia.ch/~felix/My_papers/phd/node3.html -------------------------------JOURNAL PAPERS-------------------------- F. A. Gers, J. Schmidhuber, and F. Cummins. Learning to forget: Continual Prediction with LSTM. Neural Computation, 2000. http://www.idsia.ch/~felix/My_papers/FgGates-NC.ps.gz http://www.idsia.ch/~felix/My_papers/FgGates-NC.pdf Abstract. Long Short-Term Memory (LSTM, Hochreiter & Schmidhuber, 1997) can solve numerous tasks not solvable by previous learning algorithms for recurrent neural networks (RNNs). We identify a weakness of LSTM networks processing continual input streams that are not a priori segmented into subsequences with explicitly marked ends at which the network's internal state could be reset. Without resets, the state may grow indefinitely and eventually cause the network to break down. Our remedy is a novel, adaptive "forget gate" that enables an LSTM cell to learn to reset itself at appropriate times, thus releasing internal resources. We review illustrative benchmark problems on which standard LSTM outperforms other RNN algorithms. All algorithms (including LSTM) fail to solve continual versions of these problems. LSTM with forget gates, however, easily solves them in an elegant way. --- F. A. Gers and J. Schmidhuber. LSTM recurrent networks learn simple context free and context sensitive languages. IEEE Transactions on Neural Networks, 2001. http://www.idsia.ch/~felix/My_papers/L-IEEE.ps.gz http://www.idsia.ch/~felix/My_papers/L-IEEE.pdf Abstract. Previous work on learning regular languages from exemplary training sequences showed that Long Short-Term Memory (LSTM) outperforms traditional recurrent neural networks (RNNs). Here we demonstrate LSTM's superior performance on context free language (CFL) benchmarks for recurrent neural networks (RNNs), and show that it works even better than previous hardwired or highly specialized architectures. To the best of our knowledge, LSTM variants are also the first RNNs to learn a simple context sensitive language (CSL), namely a^n b^n c^n. -------------------------------OTHER PAPERS---------------------------- Numerous additional LSTM conference papers and TRs available at: http://www.idsia.ch/~felix/Publications.html -------------------------------LSTM CODE------------------------------- C++ and Matlab code of the LSTM algorithm available at: http://www.idsia.ch/~felix/SourceCode_Data.html -------------------------------PhD POSITION---------------------------- New LSTM PhD position at IDSIA: http://www.idsia.ch/~juergen/phd2001.html ----------------------------------------------------------------------- From maass at igi.tu-graz.ac.at Fri Sep 14 12:59:49 2001 From: maass at igi.tu-graz.ac.at (Wolfgang Maass) Date: Fri, 14 Sep 2001 18:59:49 +0200 Subject: computing on spike trains Message-ID: <3BA23785.8B332C66@igi.tu-graz.ac.at> A preprint of the following paper is now online available: REAL-TIME COMPUTING WITHOUT STABLE STATES: A NEW FRAMEWORK FOR NEURAL COMPUTATION BASED ON PERTURBATIONS by Wolfgang Maass, Thomas Natschlger, and Henry Markram. (Graz Univ. of Technology, Austria, and Weizmann Institute, Israel) ABSTRACT: This paper has the goal to establish a theoretical framework for computations on spike trains that can also be applied to biologically realistic models for recurrent circuits of spiking neurons. This new theoretical framework, the liquid state machine, differs strongly from the computational models that have emerged from computer science and artificial neural networks: it is not based on transitions between stable internal states or attractors, but rather exploits the natural transient dynamics of recurrent neural circuits as a potentially powerful analog memory device. It directs attention to the investigation of trajectories of transient internal states in very high dimensional dynamical systems, thereby providing a complement to the analysis of attractors in low dimensional dynamical systems that have so far been used as primary sources of inspiration for understanding the dynamics of neural computation. Like the Turing machine this model allows for basically unlimited computational power under idealized conditions, but for real-time computing on time-varying inputs with fading memory (rather than for offline-computing on static discrete inputs like the Turing machine). Based on this new framework we have for the first time been able to carry out complex real-time computations on spike trains with biologically realistic computer models of neural microcircuits. This approach also suggests a radically new approach towards neuromorphic engineering: Look directly for efficient hardware implementations of adaptive liquid state machines in order to build devices for real-time processing of sensory inputs that capture aspects of the organisation of neural computation. Learning issues in the context of this model (especially biologically plausible algorithms for unsupervised learning and applications of reinforcement learning) are topics of current research. ------------------------------------------------------------------------------------ This paper is online available (PDF, 243 KB) as # 130 from http://www.igi.tugraz.at/maass/publications.html From xwu at gauss.Mines.EDU Sun Sep 16 11:49:41 2001 From: xwu at gauss.Mines.EDU (Xindong Wu) Date: Sun, 16 Sep 2001 09:49:41 -0600 (MDT) Subject: Knowledge and Information Systems: 3(3) and 3(4), 2001 Message-ID: <200109161549.JAA08567@gauss.Mines.EDU> Knowledge and Information Systems: An International Journal ----------------------------------------------------------- ISSN: 0219-1377 (printed version) ISSN: 0219-3116 (electronic version) by Springer-Verlag Home Page: http://www.cs.uvm.edu/~xwu/kais.html =============================================== I. Volume 3, Number 3 (August 2001) ----------------------------------- http://link.springer-ny.com/link/service/journals/10115/tocs/t1003003.htm Regular Papers - Eamonn Keogh, Kaushik Chakrabarti, Michael Pazzani, Sharad Mehrotra: Dimensionality Reduction for Fast Similarity Search in Large Time Series Databases, 263-286 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030263.htm - Juan C. Augusto, Guillermo R. Simari: Temporal Defeasible Reasoning, 287-318 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030287.htm - Mei-Ling Shyu, Shu-Ching Chen, R. L. Kashyap: Generalized Affinity-Based Association Rule Mining for Multimedia Database Queries, 319-337 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030319.htm - Guanling Lee, K. L. Lee, Arbee L. P. Chen: Efficient Graph-Based Algorithms for Discovering and Maintaining Association Rules in Large Databases, 338-355 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030338.htm - Yong S. Choi: Discovering Text Databases with Neural Nets, 356-373 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030356.htm Short Papers - Ujjwal Maulik, Sanghamitra Bandyopadhyay, John C. Trinder: SAFE: An Efficient Feature Extraction Technique, 374-387 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030374.htm Announcements - Peter Debye Prize 2002 on Knowledge Engineering II. Volume 3, Number 4 (November 2001) -------------------------------------- http://www.cs.uvm.edu/~xwu/kais/Vol-3-4.shtml Selected and Revised Papers from KDD-2000 Workshop on Distributed and Parallel Knowledge Discovery - Distributed Web Log Mining Using Maximal Large Itemsets by Mehmet Sayal and Peter Scheuermann - Parallel and Sequential Algorithms for Data Mining Using Inductive Logic by David B. Skillicorn and Yu Wang - Distributed Clustering Using Collective Principal Component Analysis by Hillol Kargupta, Weiyun Huang, Krishnamoorthy Sivakumar, and Erik Johnson - Cost Complexity-Based Pruning of Ensemble Classifiers by Andreas L. Prodromidis and Salvatore J. Stolfo Regular Papers - Arbitrating Among Competing Classifiers Using Learned Referees by Julio Ortega, Moshe Koppel, and Shlomo Argamon-Engelson - Multivariate Discretization for Set Mining by Stephen D. Bay Call for Papers - ICDM '02: The 2002 IEEE International Conference on Data Mining (pending approval), Maebashi TERRSA, Maebashi City, Japan, November 26 - 29, 2002 2001 KAIS Reviewers Author Index From becker at meitner.psychology.mcmaster.ca Mon Sep 17 10:54:47 2001 From: becker at meitner.psychology.mcmaster.ca (Sue Becker) Date: Mon, 17 Sep 2001 10:54:47 -0400 (EDT) Subject: postdoctoral positions in neural computation Message-ID: Dear all, I would like to announce two postdoctoral training opportunities, described below. The first has more of a computational neuroscience focus: to explore neuromodulator effects in learning and memory. The second has more of a neuro-signal-processing focus: to develop artificial hearing aid technology. Can you kindly bring these to the attention of any suitable candidates. I apologize if you receive multiple copies of this posting. cheers, Sue -- Sue Becker, Associate Professor Department of Psychology, McMaster University becker at mcmaster.ca 1280 Main Street West, Hamilton, Ont. L8S 4K1 Fax: (905)529-6225 www.science.mcmaster.ca/Psychology/sb.html Tel: 525-9140 ext. 23020 --------------------------------------------------------------------------- COMPUTATIONAL AND BEHAVIOURAL NEUROSCIENCE POSTDOCTORAL POSITION A unique, multi-disciplinary postdoctoral training opportunity is available to investigate the action of neurotransmitters using computational and animal models. Topics of interest include the neuromodulatory actions of dopamine in motivated behaviours, development of fears and paranoias in hyperdopaminergic conditions, learning aversive and emotional conditioned responses, and the biological bases of emotional memory formation in structures including the hippocampus and amygdala. The candidate must have a PhD in cognitive science, computer science, or a related discipline, expertise in neural network modelling, and an interest in running and overseeing learning and memory experiments involving behavioural pharmacology. Experience with animal experimentation would be an asset but is not essential, as training will be provided. Depending upon the interests of the candidate, opportunities also exist to acquire training in human functional neuroimaging, and conduct studies with clinical populations. The position is available for a minimum of two years. This research is part of a collaborative effort involving Dr. S. Becker, Department of Psychology, McMaster University (computational neuroscience), Dr. S. Kapur, Centre for Addiction and Mental Health (CAMH) and Department of Psychiatry, University of Toronto (behavioural pharmacology, human neuroimaging with PET and fMRI) and Dr. P. Fletcher, Department of Psychology, University of Toronto, and CAMH (animal models and behavioural pharmacological studies). This position will be based at both Mcmaster University in Hamilton and the CAMH in Toronto. Interested candidates should send a letter of intention, a CV and two letters of recommendation to Dr. S. Becker at the address below. Dr. Sue Becker Department of Psychology McMaster University 1280 Main Street West, Hamilton, Ont. L8S 4K1 becker at mcmaster.ca Fax: (905)529-6225 --------------------------------------------------------------------------- POST-DOCTORAL POSITIONS IN NEURO-SIGNAL PROCESSING Funding for one or more post-doctoral fellows is available to develop and test compensation algorithms for intelligent hearing aid technology. The end product will be a wearable computing device that goes well beyond the current state of the art in hearing aid design. Topics under investigation include: beamforming algorithms, noise cancellation, modelling normal and impaired cochlear filtering, cortical feedback in auditory processing, temporal processing and auditory streaming, feedback cancellation, and binaural mechanisms. A group of researchers at McMaster University received funding from NSERC, Canada for this exciting project which is being conducted in collaboration with Gennum Corporation, one of the world's largest hearing aid manufacturers. The research team, headed by Simon Haykin in Electrical and Computer Engineering (ECE), also includes Ian Bruce, currently at Johns Hopkins University and joining the McMaster ECE faculty in Jan/2002, and Sue Becker, Ron Racine, John Platt and Laurel Trainor who are faculty members in the Psychology Department, as well as several graduate students. The team's expertise spans neural modelling, signal processing, cochlear implants, neurophysiology, neural plasticity, and auditory psychophysics and neuroimaging. Preference will be given to applicants with expertise in neurobiological modelling of the auditory system or adaptive filter design or a related field. Excellent computer programming skills are essential. Please send applications, with two references, to: Prof. Simon Haykin Communications Research Laboratory McMaster Univeristy Hamilton, Ontario, Canada L8S 4K1 haykin at mcmaster.ca --------------------------------------------------------------------------- From terry at salk.edu Mon Sep 17 15:11:23 2001 From: terry at salk.edu (Terry Sejnowski) Date: Mon, 17 Sep 2001 12:11:23 -0700 (PDT) Subject: NEURAL COMPUTATION 13:10 Message-ID: <200109171911.f8HJBNv31847@purkinje.salk.edu> Neural Computation - Contents - Volume 13, Number 10 - October 1, 2001 ARTICLE Correctness of Belief Propagation in Gaussian Graphical Models of Arbitrary Topology Yair Weiss and William T. Freeman LETTERS MOSAIC Model for Sensorimotor Learning and Control Masahiko Haruno, Daniel M. Wolpert, and Mitsuo Kawato Robust Full Bayesian Learning for Radial Basis Networks Christophe Andrieu, Nando de Freitas, and Arnaud Doucet Adaptive Algorithm for Blind Separation From Noisy Time-Varying Mixtures V. Koivunen, M. Enescu, and E. Oja Evaluating Auditory Performance Limits: I. One-Parameter Discrimination Using a Computational Model for the Auditory Nerve Michael G. Heinz, H. Steven Colburn, and Laurel H. Carney Evaluating Auditory Performance Limits: II. One-Parameter Discrimination with Random-Level Variation Michael G. Heinz, H. Steven Colburn, and Laurel H. Carney Analysis and Neuronal Modeling of the Nonlinear Characteristics of a Local Cardiac Reflex in the Rat Rajanikanth Vadigepalli, Francis J. Doyle III, and James S. Schwaber Spike-Timing-Dependent Hebbian Plasticity as Temporal Difference Learning Rajesh P. N. Rao and Terrence J. Sejnowski ----- ON-LINE - http://neco.mitpress.org/ SUBSCRIPTIONS - 2001 - VOLUME 13 - 12 ISSUES USA Canada* Other Countries Student/Retired $60 $64.20 $108 Individual $88 $94.16 $136 Institution $460 $492.20 $508 * includes 7% GST MIT Press Journals, 5 Cambridge Center, Cambridge, MA 02142-9902. Tel: (617) 253-2889 FAX: (617) 577-1545 journals-orders at mit.edu ----- From mrevow at microsoft.com Mon Sep 17 12:25:28 2001 From: mrevow at microsoft.com (Michael Revow) Date: Mon, 17 Sep 2001 09:25:28 -0700 Subject: Software developer/researcher in handwriting recognition Message-ID: Microsoft Handwriting Recognition Group Software Developer/Researcher Responsibilities center on incorporating research into the Microsoft Handwriting Recognition (HR) engine. Experience in handwriting recognition technology is not required, but a solid background in Computer Science, Mathematics, or Statistics is needed, as is the desire and ability to learn new techniques from textbooks or research papers. The HR engine is being developed for a variety of languages and includes components that implement neural nets, hidden Markov models, signal processing, statistical search, n-gram statistical language modeling, and context-free grammar language modeling. Improvements to the HR engine are targeted at increasing performance and accuracy, and improving usability problems for mainstream use in applications, such as note taking into Microsoft Office or the TabletPC. Qualifications required include at least a Bachelors degree in Computer Science, Mathematics, EE/Signal Processing, or related field. An advanced degree would be desirable. A strong interest and ability to write and support your own code in C/C++ is crucial to being successful in the group. Interested candidates should submit a recent copy of their resume and indicate their availability to: Michael Revow, 1 Microsoft Way, Redmond WA, 98052, USA. or by email to: mrevow at microsoft.com. From ken at phy.ucsf.edu Tue Sep 18 01:37:47 2001 From: ken at phy.ucsf.edu (Ken Miller) Date: Mon, 17 Sep 2001 22:37:47 -0700 Subject: Paper available: Processing in Layer 4 of the Neocortical Circuit Message-ID: <15270.56747.854955.753714@coltrane.ucsf.edu> The following review paper is available from http://www.keck.ucsf.edu/~ken (click on 'publications'): Miller, K.D., D.J. Simons and D.J. Pinto (2001). ``Processing in Layer 4 of the Neocortical Circuit: New Insights From Visual and Somatosensory Cortex''. Current Opinion in Neurobiology 11, 488-497. Summary: Recent experimental and theoretical results in cat primary visual cortex (V1) and in the whisker-barrel fields of rodent primary somatosensory cortex (S1) suggest common organizing principles for layer 4, the primary recipient of sensory input from thalamus. Response tuning of layer 4 cells is largely determined by a local interplay of feedforward excitation (from thalamus) and feedforward inhibition (from layer 4 inhibitory interneurons driven by thalamus). Feedforward inhibition dominates excitation, inherits its tuning from the thalamic input, and sharpens the tuning of excitatory cells. Recurrent excitation enhances responses to effective stimuli. We review results leading to these common pictures, and also highlight remaining differences between the two systems. Ken Kenneth D. Miller telephone: (415) 476-8217 Associate Professor fax: (415) 476-4929 Dept. of Physiology, UCSF internet: ken at phy.ucsf.edu 513 Parnassus www: http://www.keck.ucsf.edu/~ken San Francisco, CA 94143-0444 From omlin at cs.sun.ac.za Tue Sep 18 09:14:34 2001 From: omlin at cs.sun.ac.za (Dr Christian W Omlin) Date: Tue, 18 Sep 2001 15:14:34 +0200 (SAST) Subject: Postdoctoral Positions Message-ID: The Department of Computer Science at the University of the West- ern Cape (South Africa) has identified the area of intelligent systems as one of its main research thrusts. We aim to position ourselves to assume a national leadership role in this emerging field. In order to further strengthen our research capacity, we invite applications for 2 POSTDOCTORAL POSITIONS for an initial 1-year period in our Intelligent Systems Group. The appointment can be extended by an additional 2 years. Areas of interest include but are not restricted to neural networks signal processing internet applications data mining intelligent networks intelligent agents computational learning theory For applicants with research interests in bioinformatics, there exist opportunities for collaboration with the South African Na- tional Bioinformatics Institute. The successful candidates may either pursue their own research programmes or may participate in existing projects. No teaching duties are expected from the postdoctoral research fellows; however, there exists the opportunity to supervise MSc/PhD students or to give a postgraduate course if he/she so desires. We offer a competitive, tax-free stipend, a friendly atmosphere, and state-of-the-art equipment. We encourage publication of re- search results and offer financial support for travel to confer- ences. The University of the Western Cape is located 20 minutes from the center of cosmopolitan Cape Town which features scenic views, sandy beaches, great opportunities for outdoor activities, na- tional parks withs stunnning fauna and flora, and plenty of arts and culture. We also have one of the most pleasant climates in the world with 300 days of sunshine and very mild winters. We are also located 20 minutes from the heart of South Africa's wineland where world-class wines are produced at numerous vinyards. Please send your CV, a list of publications, copies of two repre- sentative papers and a statement of your research interests to comlin at uwc.ac.za. Also, please have 3 referees send letters of recommendation to the same address. Deadline for applications is November 30, 2001. ----------------------------------------------------------------- Prof. Christian W. Omlin Phone: +27-21-959/3010/2967 Intelligent Systems Group Fax: +27-21-959-2577 Department of Computer Science E-mail: comlin at uwc.ac.za University of the Western Cape URL: http://www.cs.uwc.ac.za Bellville 7535 South Africa ----------------------------------------------------------------- From cindy at cns.bu.edu Tue Sep 18 14:48:20 2001 From: cindy at cns.bu.edu (Cynthia Bradford) Date: Tue, 18 Sep 2001 14:48:20 -0400 Subject: Neural Networks 14(8) Message-ID: <200109181848.OAA28253@retina.bu.edu> NEURAL NETWORKS 14(8) Contents - Volume 14, Number 8 - 2001 ------------------------------------------------------------------ NEURAL NETWORKS LETTER: Global exponential stability of delayed Hopfield neural networks Tianping Chen CONTRIBUTED ARTICLES: ***** Psychology and Cognitive Science ***** A neurodynamical model for selective visual attention using oscillators Silvia Corchs and Gustavo Deco ***** Neuroscience and Neuropsychology ***** Information processing in dendrites, I: Input pattern generalization Kevin N. Gurney Information processing in dendrites, II: Information theoretic complexity Kevin N. Gurney ***** Mathematical and Computational Analysis ***** A modified general regression neural network (MGRNN) with new, efficient training algorithms as a robust "black box" tool for data analysis Dirk Tomandl and Andreas Schober An approach to guaranteeing generalization in neural networks J. Gary Polhill and Michael K. Weir Algebraic geometrical methods for hierarchical learning machines Sumio Watanabe A hybrid learning network for shift-invariant recognition Ruye Wang A dynamical model for the analysis and acceleration of learning in feedforward networks Nikolaos Ampazis, Stavros J. Perantonis, and John G. Taylor Estimates of average complexity of neurocontrol algorithms Tomas Hrycej Fuzzylot: A novel self-organizing fuzzy-neural rule-based pilot system for automated vehicles Michel Pasquier, Chai Quek, and Mary Toh ***** Engineering and Design ***** Real time distributed processing of multiple associated pulse pattern sequences A.J.B. Travis Myopotential denoising of ECG signals using wavelet thresholding methods Vladimir Cherkassky and Steven Kilts ------------------------------------------------------------------ Electronic access: www.elsevier.com/locate/neunet/. Individuals can look up instructions, aims & scope, see news, tables of contents, etc. Those who are at institutions which subscribe to Neural Networks get access to full article text as part of the institutional subscription. Sample copies can be requested for free and back issues can be ordered through the Elsevier customer support offices: nlinfo-f at elsevier.nl usinfo-f at elsevier.com or info at elsevier.co.jp ------------------------------ INNS/ENNS/JNNS Membership includes a subscription to Neural Networks: The International (INNS), European (ENNS), and Japanese (JNNS) Neural Network Societies are associations of scientists, engineers, students, and others seeking to learn about and advance the understanding of the modeling of behavioral and brain processes, and the application of neural modeling concepts to technological problems. Membership in any of the societies includes a subscription to Neural Networks, the official journal of the societies. Application forms should be sent to all the societies you want to apply to (for example, one as a member with subscription and the other one or two as a member without subscription). The JNNS does not accept credit cards or checks; to apply to the JNNS, send in the application form and wait for instructions about remitting payment. The ENNS accepts bank orders in Swedish Crowns (SEK) or credit cards. The INNS does not invoice for payment. ---------------------------------------------------------------------------- Membership Type INNS ENNS JNNS ---------------------------------------------------------------------------- membership with $80 (regular) SEK 660 (regular) Y 13,000 (regular) Neural Networks (plus 2,000 enrollment fee) $20 (student) SEK 460 (student) Y 11,000 (student) (plus 2,000 enrollment fee) ----------------------------------------------------------------------------- membership without $30 SEK 200 not available to Neural Networks non-students (subscribe through another society) Y 5,000 (student) (plus 2,000 enrollment fee) ----------------------------------------------------------------------------- Name: _____________________________________ Title: _____________________________________ Address: _____________________________________ _____________________________________ _____________________________________ Phone: _____________________________________ Fax: _____________________________________ Email: _____________________________________ Payment: [ ] Check or money order enclosed, payable to INNS or ENNS OR [ ] Charge my VISA or MasterCard card number ____________________________ expiration date ________________________ INNS Membership 19 Mantua Road Mount Royal NJ 08061 USA 856 423 0162 (phone) 856 423 3420 (fax) innshq at talley.com http://www.inns.org ENNS Membership University of Skovde P.O. Box 408 531 28 Skovde Sweden 46 500 44 83 37 (phone) 46 500 44 83 99 (fax) enns at ida.his.se http://www.his.se/ida/enns JNNS Membership c/o Professor Takashi Nagano Faculty of Engineering Hosei University 3-7-2, Kajinocho, Koganei-shi Tokyo 184-8584 Japan 81 42 387 6350 (phone and fax) jnns at k.hosei.ac.jp http://jnns.inf.eng.tamagawa.ac.jp/home-j.html ----------------------------------------------------------------- From taketani at tensorbio.com Tue Sep 18 18:34:50 2001 From: taketani at tensorbio.com (Makoto Taketani) Date: Tue, 18 Sep 2001 15:34:50 -0700 Subject: Director of Physiology Position Available at Tensor Biosciences Message-ID: <002201c14092$21fcc0e0$0300a8c0@CORP.TENSORBIO.COM> Director of Physiology Position Tensor Biosciences Irvine, CA USA Area of expertise: Slice Physiology Tensor Biosciences is looking for a high-level slice physiologist as a Director of Physiology. The new Director of Physiology will manage Tensor's acute physiology group which currently has four members and is expected to grow further during the next year. This position will involve managing both research and contract work. In addition, the candidate will be asked to assist with various business development activities, and as the company grows, he or she will be responsible for growing their team into a department. A strong candidate for this position should have a doctorate in the biological sciences and the following characteristics: (1) an enthusiastic start-up attitude, that is, they must be very flexible and willing to work long and hard at a variety of jobs with eye towards increasing the value of their stock; (2) strong communication skills as demonstrated by the candidates writing, presentation, and teaching experience, with any business development experience being a big plus; (3) strong management skills, with an emphasis on team building; and (4) a strong background in acute brain slice physiology with pharmaceutical industry experience being a big plus. Obviously Tensor is looking for an exceptional individual to fill this opening. Tensor will pay a salary at the upper end of the industry range for such a person, and provide he or she with a very generous stock option package. In addition, the person who fills this job will have ample opportunity to write patents and papers, and develop high-level pharmaceutical industry contacts. Tensor Biosciences (www.tensorbio.com) is developing a revolutionary new kind of drug discovery technolgies based upon Panasonic's patented multi-electrode array hardware (www.med64.com) and advanced informatics software being developed in-house to differentiate drugs and predict their effects. This work is proceeding in collaboration with a laboratory of professor Gary Lynch of the University of California at Irvine. Tensor is already delivering on its first drug analysis contract and early signs suggest that its sales of testing services and the value of the company will grow very rapidly in the next year. To apply, send a letter of application and CV to Tensor at the address listed below. Applications will be considered as they are received. Contact: Heidi Merrill Human Resources Tensor Biosciences 101 Theory, Suite 250 Irvine, CA 92697 Office:(949) 258-0309 Fax:(949) 258-0321 heidimerrill at tensorbio.com ============================================= Makoto Taketani, Ph.D. Technology Development Center Matsushita Electric Corporation of America Tel: 949-258-0310; FAX: 949-258-0321 Net: taketani at med64.com http://www.med64.com ------------------------------------------ From Chris.Diehl at jhuapl.edu Tue Sep 18 09:23:31 2001 From: Chris.Diehl at jhuapl.edu (Diehl, Chris P.) Date: Tue, 18 Sep 2001 09:23:31 -0400 Subject: Ph.D. Thesis Announcement Message-ID: <91D1D51C2955D111B82B00805F1998950BD8C946@aples2.jhuapl.edu> Dear Connectionists, The following Ph.D. thesis is now available at http://www.cpdiehl.org. Toward Efficient Collaborative Classification for Distributed Video Surveillance Christopher P. Diehl Ph.D. Thesis Department of Electrical and Computer Engineering Carnegie Mellon University Abstract In this thesis, we propose a general strategy for automated video surveillance that relies on collaboration between the surveillance system and the user. Such collaboration enables the user to help the system incrementally acquire the necessary context for truly robust surveillance. The success of this strategy is dependent on the ability of the system to identify novel instances of known or unknown classes that it does not understand. This, in turn, allows the user to focus only on the observations with the highest uncertainty that require interpretation. Designing a real-time classification process that supports novelty detection is nontrivial. The real-time constraint dictates computational simplicity, whereas novelty detection requires a high dimensional feature space to aid in discriminating between the known and unknown classes. The majority of this work focuses on the problem of simultaneously satisfying these conflicting constraints. We consider these issues in the context of a relevant surveillance task and evaluate the performance of the resulting classification process in the CMU Cyberscout distributed video surveillance system. Dr. Chris Diehl System and Information Sciences Group Research and Technology Development Center Applied Physics Laboratory Johns Hopkins University 443-778-3457 (Office) 443-778-6904 (Fax) http://www.cpdiehl.org From cohn+jmlr at cs.cmu.edu Wed Sep 19 15:48:34 2001 From: cohn+jmlr at cs.cmu.edu (JMLR) Date: Wed, 19 Sep 2001 15:48:34 -0400 Subject: Two new papers in the Journal of Machine Learning Research Message-ID: The Journal of Machine Learning Research (www.jmlr.org) is pleased to announce the availability of two new papers in electronic form: Tracking the Best Linear Predictor Mark Herbster and Manfred K. Warmuth Journal of Machine Learning Research 1(Sep):281-309, 2001 Prior Knowledge and Preferential Structures in Gradient Descent Learning Algorithms Robert E. Mahony, Robert C. Williamson Journal of Machine Learning Research 1(Sep):311-355, 2001 ---------------------------------------- Tracking the Best Linear Predictor Mark Herbster and Manfred K. Warmuth Journal of Machine Learning Research 1(Sep):281-309, 2001 Abstract In most on-line learning research the total on-line loss of the algorithm is compared to the total loss of the best off-line predictor u from a comparison class of predictors. We call such bounds static bounds. The interesting feature of these bounds is that they hold for an arbitrary sequence of examples. Recently some work has been done where the predictor u_t at each trial t is allowed to change with time, and the total on-line loss of the algorithm is compared to the sum of the losses of u_t at each trial plus the total ``cost'' for shifting to successive predictors. This is to model situations in which the examples change over time, and different predictors from the comparison class are best for different segments of the sequence of examples. We call such bounds shifting bounds. They hold for arbitrary sequences of examples and arbitrary sequences of predictors. Naturally shifting bounds are much harder to prove. The only known bounds are for the case when the comparison class consists of a sequences of experts or boolean disjunctions. In this paper we develop the methodology for lifting known static bounds to the shifting case. In particular we obtain bounds when the comparison class consists of linear neurons (linear combinations of experts). Our essential technique is to project the hypothesis of the static algorithm at the end of each trial into a suitably chosen convex region. This keeps the hypothesis of the algorithm well-behaved and the static bounds can be converted to shifting bounds. ---------------------------------------- Prior Knowledge and Preferential Structures in Gradient Descent Learning Algorithms Robert E. Mahony, Robert C. Williamson Journal of Machine Learning Research 1(Sep):311-355, 2001 Abstract A family of gradient descent algorithms for learning linear functions in an online setting is considered. The family includes the classical LMS algorithm as well as new variants such as the Exponentiated Gradient (EG) algorithm due to Kivinen and Warmuth. The algorithms are based on prior distributions defined on the weight space. Techniques from differential geometry are used to develop the algorithms as gradient descent iterations with respect to the natural gradient in the Riemannian structure induced by the prior distribution. The proposed framework subsumes the notion of "link-functions". ---------------------------------------- These papers and earlier papers in Volume 1 are available electronically at http://www.jmlr.org in PostScript, PDF and HTML formats; a bound, hardcopy edition of Volume 1 will be available later this year. -David Cohn, Managing Editor, Journal of Machine Learning Research ------- This message has been sent to the mailing list "jmlr-announce at ai.mit.edu", which is maintained automatically by majordomo. To subscribe to the list, send mail to listserv at ai.mit.edu with the line "subscribe jmlr-announce" in the body; to unsubscribe send email to listserv at ai.mit.edu with the line "unsubscribe jmlr-announce" in the body. From mkm at hnc.com Wed Sep 19 23:26:12 2001 From: mkm at hnc.com (McClarin, Melissa) Date: Wed, 19 Sep 2001 20:26:12 -0700 Subject: HNC Software seeks a Senior Staff Scientist Message-ID: <72A838A51366D211B3B30008C7F4D363085AE7F7@pchnc.hnc.com> HNC Software is seeking a full time Senior Staff Scientist, a job based at our headquarters in San Diego, California. We provide Customer Insight through Intelligent Response, decision management and customer analytics software that enables companies in the financial, telecommunications, e-commerce and insurance industries to acquire, manage and retain customers. Senior Staff Scientist Job Duties/Responsibilities: General Duties: PhD with some experience in the development of predictive models to real-world business applications. Knowledge in worker compensation/auto insurance. Superior programming skills. Excellent communication skills(both writing and verbal). Specific Duties: Design and implement innovative solutions to a wide variety of technical problems in insurance industry using data mining techniques, statistics and neural network modeling. As a technical lead of projects and product development, contribute to all phases of model development which including data analysis, solution design, model training, and presentation of results. Provides technical guidance to other team members. Works with other functional groups to ensure integrity of deliverable. Required Qualifications: MS/PhD in computer science, statistics, math, Engineering, or related field. Demonstrated experience applying advanced statistics, data mining techniques to real world data. Must be a strong problem solver and fast learner, possessing excellent analytical abilities. Ability to work independently and provides team technical guidance. Comfortable with a variety statistical packages and computer systems, such as SAS, C, Unix, Windows. Preferred Qualifications: PhD with 5 years experience in the development of predictive models to real-world business applications. Knowledge in worker compensation/auto insurance, and software development. Superior programming skills. Excellent communication skills(both writing and verbal). Experience as a team leader To apply, please use one of the following methods: Mail: 5935 Cornerstone Ct. W San Diego, CA 92121 Email: hireme at hnc.com Online: http://www.hnc.com/careers_12/index - job number 366 From schrott at in.tum.de Thu Sep 20 03:50:46 2001 From: schrott at in.tum.de (Dr. Gerhard Schrott) Date: Thu, 20 Sep 2001 09:50:46 +0200 Subject: Research or Post-Doc Research Position at TU Munich Message-ID: <3BA99FD6.F99F24C0@in.tum.de> Technical University of Munich Department of Computer Science The Department of Computer Science, Robotics and Embedded Systems unit, invites applications for a Research or Post-Doc Research Position within the EU-funded long-term research programme "Neuroinformatics for Living Artefacts" in a pan-European research project aiming at exploring the fundamentals and the potential of Imitation Learning for future robot systems. Applications would be particularly welcome from candidates with research interests (and a record of innovative research results) in one or more of the following areas: robotics (e.g. Sensor data processing and sensor data fusion; Robust vision/tracking for robots; Learning and adaptive control; Behaviour engineering and architectures) and/or neuro-modelling (e.g. modelling of dynamical systems; Optimization and nonlinear dynamics; Coupling of motor behaviour and visual perception; pattern generation for motion and postural control; Transfer of the mathematical concepts into algorithms). The successful candidate will have a graduate or a postgraduate qualification - ideally to doctoral level - in computer science or neurosciences, together with practical experience in robotics. Further details may be obtained from Prof. A. Knoll, knoll at in.tum.de. Closing date: 5 October 2001. Applications including the CV, certificates, publication list should be sent to: Technische Universitaet Muenchen, Fakultaet fuer Informatik, LS VI - Robotics and Embedded Systems, z.H. R. te Vehne, Orleansstrasse 34, 80667 Muenchen, Germany, or by E-mail to Vehne at in.tum.de From bengio at idiap.ch Fri Sep 21 07:57:49 2001 From: bengio at idiap.ch (Samy Bengio) Date: Fri, 21 Sep 2001 13:57:49 +0200 (MEST) Subject: Torch: a new machine learning library in C++/GPL Message-ID: Dear connectionists people, We would like to announce the availability of yet another machine learning library, written in C++ under the GPL license. This new library is named "Torch" and has been written mainly by Ronan Collobert, also known for his widely used SVMTorch package for SVMs. Other contributors include Samy Bengio and Johnny Mari=E9thoz. Currently, the main features of Torch are the following: - A lot of things in gradient-machines, that is, machines which could be learned with gradient descent. This includes Multi-Layered Perceptrons, Radial Basis Functions and Mixtures of Experts. In fact there are a lot of small "modules" available (Linear, Tanh, SoftMax...) that you can plug as you want to get what you want. - Support Vector Machine, in classification and regression (mostly the same code as in SVMTorch but now integrated in the Torch library). - A Distribution package which includes for the moment Kmeans, Gaussian Mixture Models (GMMs) and Hidden Markov Models (HMMs). - A few non-parametric models such as K-nearest-neighbors, Parzen Regression and Parzen Density Estimator. - Tools to do either "train/test" experiments or K-fold cross-validation - A lot of measurers (to print for instance the mean squared error or the classification error of as many datasets as wanted, during training or testing). - A few typical "main.cc" examples to understand how to create your own experiments such as mlp.cc (train and test an MLP), gmm.cc (maximize the likelihood of a GMM), hmm.cc (maximize the likelihood of an HMM), mixture_softmax (train and test a mixture of experts). This library is intented to provide the state-of-the-art of the best algorithms in machine learning. Therefore, if you know C++, are working in machine learning, and want to develop your own algorithms or use well-known machine learning algorithms, Torch is for you! Visit the official Torch website at http://www.torch.ch to know more about Torch and/or download it. Note that, of course, Torch is and will always be under development... Note also that it was designed for Unix and Linux systems... Have fun! ----- Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From crocker at CoLi.Uni-SB.DE Tue Sep 25 04:31:05 2001 From: crocker at CoLi.Uni-SB.DE (Matthew Crocker) Date: Tue, 25 Sep 2001 10:31:05 +0200 Subject: Positions in Language Processing In-Reply-To: <200108311318.PAA07149@top.coli.uni-sb.de> Message-ID: <200109250834.KAA07734@top.coli.uni-sb.de> Two Research Positions Available in Computational and Experimental Psycholinguistics, Saarbruecken, Germany The Department of Computational Linguistics, Saarland University is seeking to fill two research positions in the areas of Computational and Experimental Psycholinguistics. Persons taking up the positions will be involved in a newly funded project entitled: "Adaptive Mechanisms for Human Language Processing". The project aims to develop wide-coverage, probabilistic models of human language processing, as informed by evidence obtained from large corpora, and both on-line and off-line psycholinguistic experiments. 1. Computational Linguist: experience in the development of probabilistic models of language processing. In particular, we are interested in developing incremental models of syntactic and semantic processing. Knowledge of one or more of the following is desirable: probabilistic parsing techniques, machine learning of natural language, connectionist language modelling, and experience working with large corpora. 2. Experimental Psycholinguist: experience in designing, running and analysing psycholinguistic experiments, and familiarity with (some of) the following paradigms: self-paced reading, eye-tracking (fixed or head-mounted), language production, and web-based experiments. Candidates should have research experience in a relevant subject area, and ideally will hold a PhD. The position is on the BAT IIa scale (roughly up to DM 80K per annum, depending on age and family status) and is tenable for 3 years with the possibility of renewal. The positions are available from January 1st, 2002. Applications received before November 1, 2001 are assured fullest consideration. Interested persons should send a letter of application giving contact details for three possible referees and a full CV to (e-mail applications are also welcome): Prof. Dr. Matthew W. Crocker Psycholinguistics Group, Gebaeude 17 Department of Computational Linguistics Saarland University 66041 Saarbruecken, Germany E-mail: crocker at coli.uni-sb.de The Department offers state of the art research facilities including head-mounted and DPI eye-tracking equipment, powerful Unix servers for statistical modeling, and an extensive corpus infrastructure. Saarland University has an international profile in computational linguistics, cognitive science and computer science. Research is supported by a European Graduate School for "Language Technology and Cognitive Systems" (joint with Edinburgh University) and a Centre of Excellence (SFB 378) in the area of "Resource Adaptive Cognitive Processes". The University of Saarland seeks to increase the proportion of women in positions where they are under-represented, and therefore particularly encourages applications from women. In the selection procedure, disabled persons with equivalent qualifications will be favoured. From: Martin Giese Subject: PhD position in computer vision / biomedical engineering From bogus@does.not.exist.com Wed Sep 26 17:39:05 2001 From: bogus@does.not.exist.com () Date: Wed, 26 Sep 2001 23:39:05 +0200 Subject: No subject Message-ID: The Laboratory for Action Representation and Learning at the Department of Cognitive Neurology and the Max-Planck Institute for Biological Cybernetics in Tuebingen (Germany) offers a PhD position for students in engineering, computer science, physics or mathematics. Aim of the project is the development of technical methods for the analysis of complex movements of neurological patients. On the theory side, the project focuses on the development of new learning techniques for the representation of complex movements. The practical side of the project includes data acquisition with modern motion capture systems, and the analysis and visualization of the data using methods from computer vision and computer graphics. In addition, the successful candidate should be willing to communicate with neurologists and patients. The project will be realized in close collaboration with the Department for Neurology at the Universitaetsklinik Tuebingen. An interesting environment for computer vision and machine learning is provided by the divisions for Virtual Reality and machine learning reasearch at the Max-Planck-Institute. In addition, the group has close contacts with the Center for Biological and Computational Learning at M.I.T. The ideal candidate has a good computational background, and some experience in computer vision / graphics, machine learning, or robotics. Payment will be, dependent on prior experience, BAT IIa / 2 or BAT IIa for 3 years (extendable). Starting date should be as soon as posible. The position is funded by the German Volkswagen Foundation. For further information please contact: Dr. Martin Giese Laboratory for Action Representation and Learning Max-Planck-Institut for Biological Cybernetics Spemannstr. 34 D-72076 Tuebingen GERMANY Email: martin.giese at tuebingen.mpg.de Applicants are asked to submit their CV, a bibliography, and the names of two references. Applications and references should be sent by email to the same address. -------------------------------------------------------- Dr. Martin Giese Laboratory for Action Representation and Learning Max-Planck-Institut for Biological Cybernetics Spemannstr. 34 D-72076 Tuebingen GERMANY Tel.: +49 7071 601 724 Fax: +49 7071 601 616 Email: martin.giese at tuebingen.mpg.de -------------------------------------------------------- From murphyk at cs.berkeley.edu Wed Sep 26 22:29:18 2001 From: murphyk at cs.berkeley.edu (Kevin Murphy) Date: Wed, 26 Sep 2001 19:29:18 -0700 Subject: paper on BNT now available Message-ID: <3BB28EFE.236528D0@cs.berkeley.edu> I am pleased to announce the following paper "The Bayes Net Toolbox for Matlab", by Kevin Murphy. Published in: Computing Science and Statistics: Proceedings of Interface, volume 33, 2001 Available at http://HTTP.CS.Berkeley.EDU/~murphyk/Papers/bnt.ps.gz http://HTTP.CS.Berkeley.EDU/~murphyk/Papers/bnt.pdf Abstract: The Bayes Net Toolbox (BNT) is an open-source Matlab package for directed graphical models. BNT supports many kinds of nodes (probability distributions), exact and approximate inference, parameter and structure learning, and static and dynamic models. BNT is widely used in teaching and research: the web page has received over 28,000 hits since May 2000. In this paper, we discuss a broad spectrum of issues related to graphical models (directed and undirected), and describe, at a high-level, how BNT was designed to cope with them. We also compare BNT to other software packages for graphical models, and to the nascent OpenBayes effort. From strom at ece.ogi.edu Wed Sep 26 20:37:23 2001 From: strom at ece.ogi.edu (Dan Hammerstrom) Date: Wed, 26 Sep 2001 17:37:23 -0700 Subject: Post-Doc position Message-ID: <4.2.0.58.20010926173522.01d68ef8@spruce.ece.ogi.edu> Postdoctoral Position in Computational Modelling for Sensorimotor Control A postdoctoral training opportunity is available to investigate computational models for sensorimotor control with application to autonomous robot navigation. The research aims at developing models for bottom-up/top-down sensory integration and high-level representations for visuomotor control. The current investigations address sparse representations and on-line learning within such representations, working memory and goal memory representations and learning; representation of risk, risk anticipation and risk aversion. The candidate must have a PhD in cognitive science, computer science/engineering or a related discipline, expertise in one or more of the following: neural modelling, learning algorithms, machine vision, roving robot programming. This research is part of a collaborative effort with funding from NASA under its Revolutionary Computing Program and a Gordon & Betty Moore Research Scholarship. The research involves Prof. Marwan Jabri (OHSU), Prof. Dan Hammerstrom (OHSU) and Prof. Terrence Sejnowski (Salk). The present position is located at the OGI School of Science and Engineering, OHSU West Campus, Beaverton, Oregon. Interested candidates should send a letter of intention, a CV and two letters of recommendation to Prof. M. Jabri at the address below. Marwan Jabri OGI School of Science & Engineering Oregon Health and Sciences University 20000 N.W. Walker Rd. Beaverton, OR 97006, USA Tel: (+1-503) 748-7274 Fax: (+1-503) 748-4070 marwan at ece.ogi.edu From duch at phys.uni.torun.pl Wed Sep 26 10:41:10 2001 From: duch at phys.uni.torun.pl (Wlodzislaw Duch) Date: Wed, 26 Sep 2001 16:41:10 +0200 Subject: ICNNSC 2002, Sixth International Conference on Neural Networks and Soft Computing, call for papers Message-ID: ICNNSC 2002, http://www.icnnsc.pcz.czest.pl Sixth International Conference on Neural Networks and Soft Computing June 11-15, 2002, Zakopane, Poland organised by Polish Neural Networks Society in cooperation with IEEE Neural Networks Council. Honorary chairs Lotfi Zadeh - USA Jacek Zurada - USA General co-chairs Wlodzislaw Duch - Poland Janusz Kacprzyk - Poland Leszek Rutkowski - Poland Ryszard Tadeusiewicz - Poland Zdzislaw Pawlak - Poland The Sixth International Conference on Neural Networks and Soft Computing ICNNSC 2002 will be held in Zakopane (situated in the HighTatra mountains), Poland on 11-15 June, 2002. The conference will provide an excellent opportunity for scientist and engineers to present and discuss the latest scientific results and methods. The conference will include keynote addresses, contributed papers, and numerous lectures and tutorials on a wide range of topics. Important Dates: 1. Submission of papers in accordance with Springer guidelines (6 pages max.): February 15, 2002 2. Notification of acceptance: March 30, 2002 3. Submission of camera-ready papers May5, 2002 4. Conference date: 11-15 June, 2002 Scope The conference covers all topics in neural networks, fuzzy systems, evolutionary computation, hybrid methods, including but not limited to: Supervised and unsupervised learning Neural network theory Neural network architectures Hardware implementations Fuzzy logic Fuzzy optimisation Fuzzy control Fuzzy computing with words Theory of evolutionary algorithms Evolutionary design Evolutionary scheduling and optimisation Rough sets theory: foundations and applications Multi-agent systems Data mining Applications: prediction, identification, pattern recognition, image and signal processing, speech and computer vision, financial engineering and forecasting, medicine, industry ... The working language of the conference is English. Only original, unpublished papers in the aforementioned fields are invited. Authors should submit 3 hard copies of full papers (up to six pages) to the conference office and an electronic version to: icnnsc at kik.pcz.czest.pl before February 15, 2002. Paper Submission and Publication The papers should be organized in accordance with a common scientific structure (abstract, state of the art in the field, intention, used methodology, obtained results and references). Papers will be refereed by an international committee, and accepted on the basis of their scientific merit and relevance to the conference topics. After the notification of acceptance (March 30, 2002), authors will be allowed to make a correction in accordance with the suggestions of the reviewers and submit final camera-ready papers before May 5, 2002. Abstracts of accepted papers will be available during the conference in the form of a brochure. The conference proceedings will be published in the Springer-Verlag series "Advances in Soft Computing" and distributed among the participants after the conference. Accepted papers must be presented by author(s) personally to be published in the conference proceedings. Venue The conference is organized in Poland's premier mountain resort Zakopane, the beautiful and picturesque capital of the Polish High Tatra Mountains. Even though Zakopane is called Poland's winter capital, it is an excellent place to visit in spring. Situated at the foot of the Tatra Mountains and the Tatra National Park (with Poland's highest peak, Rysy mountain (2499m)), it offers you a possibility to hike on over 240km of marked hiking trails for beginners as well as experienced climbers and explore caves accessible only in summer. The town is filled with historic monuments, museums, art galleries and exhibitions. On your way to Zakopane you may visit Cracow, one of the most historic Polish cities, with a wonderfully preserved old city center and the largest medieval square in Europe. For details please visit the ICNNSC2002 Web page http://www.icnnsc.pcz.czest.pl From l.s.smith at cs.stir.ac.uk Thu Sep 27 04:29:15 2001 From: l.s.smith at cs.stir.ac.uk (Leslie Smith) Date: Thu, 27 Sep 2001 09:29:15 +0100 Subject: Research posts in Scotland: INCITE Message-ID: <3BB2E35B.F6CA59CB@cs.stir.ac.uk> New Institute based at the University of Stirling, Scotland, with research fellow positions in Stirling, Edinburgh and Paisley Universities, Scotland: Institute for Neuronal Computational Intelligence and Technology (INCITE) INCITE aims to design systems which combine the speed of silicon with the robustness of human performance. This will underpin the next generation of IT systems, enabling them to approach the complexity of human performance. INCITE is a newly funded Institute integrating expertise from the Universities of Stirling, Edinburgh and Paisley on neuromorphic engineering, robotics, neural networks and cognitive and computational neuroscience. INCITE will combine research and technology transfer functions. The INCITE web page is at http://www.cn.stir.ac.uk/incite/, and contact details can be found from there. Research Fellow (3 posts) INCITE will perform research at Stirling, Paisley and Edinburgh Universities. One Research Fellow will be appointed at each location, to assist with ongoing research. For more information about each location, see the www page above. The posts are for three years, starting after 1 Feb, 2002. Salary will be within the UK Universities Research & Analogous Staff Scale Grade 1A (UK?17,451-26,229 p.a.). Quote Reference 5368S/413 (Stirling), 5368E/413 (Edinburgh) and 5368P/413 (Paisley) For informal discussion prospective applicants may phone Professor W?rg?tter (+(44) 1786 466369) or Professor Smith (+(44) 1786 467435). Further particulars for all posts are available from the Personnel Office, University of Stirling, Stirling, FK9 4LA, tel: (+(44) 1786 467028, fax (+(44) 1786) 466155 or email personnel at stir.ac.uk quoting the appropriate reference number. Closing date for applications: Wednesday, 10 October 2001. www.personnel.stir.ac.uk AN EQUAL OPPORTUNITIES EMPLOYER -- Professor Leslie S. Smith, Head of Department, Dept of Computing Science and Mathematics, University of Stirling, Stirling FK9 4LA, Scotland l.s.smith at cs.stir.ac.uk Tel (44) 1786 467435 Fax (44) 1786 464551 www http://www.cs.stir.ac.uk/~lss/ From shahar at discus.anu.edu.au Thu Sep 27 23:50:48 2001 From: shahar at discus.anu.edu.au (Shahar Mendelson) Date: Fri, 28 Sep 2001 05:50:48 +0200 Subject: Machine Learning Summer School Message-ID: <4.2.2.20010928055015.00b63d00@in.zahav.net.il> Machine Learning Summer School We would like to inform you that The Australian National University will be hosting a Machine Learning Summer School. The school will consist of six short courses and a series of special invited talks, taught by experts from Australia and overseas. The school will be held between February 11 and February 22, 2002. It is suitable for all levels, both for people without previous knowledge in Machine Learning and those wishing to broaden their expertise in this area. It will allow the participants to get in touch with international experts in this field. The list of topics and speakers includes: - Reinforcement Learning Peter Bartlett, Biowulf Technologies - Boosting Ron Meir, Technion - Statistical Learning Theory Shahar Mendelson, ANU - Online Learning and Bregmann divergences Gunnar Ratsch, ANU - Support Vector Machines Bernhard Schoelkopf, Biowulf Technologies and MPIK - Bayesian Kernel Methods Alex Smola, ANU We are offering a limited number of scholarships for Australian students with a strong academic background. Students who are interested should include their CV with their application. The registration cost of the school is $1,200 per person for participants from industry and $240 per person for academics. Students are eligible for a further discount and may register for $120 per person. All prices are in Australian dollars and include GST. The closing date for early registration is December 31, 2001. Registration after this date will be subject to 33% surcharge. The closing date for scholarship applications is December 7, 2001. For further information, visit our website at http://mlg.anu.edu.au/summer2002 or send e-mail to Michelle.Moravec at anu.edu.au or Diane.Kossatz.anu.edu.au. Regards, Alex Smola and Shahar Mendelson From tleen at cse.ogi.edu Fri Sep 28 19:30:54 2001 From: tleen at cse.ogi.edu (Todd Leen) Date: Fri, 28 Sep 2001 16:30:54 -0700 Subject: Postdoctoral Position in Neural Modeling Message-ID: <3BB5082D.CE1652F1@cse.ogi.edu> Postdoctoral Research Position in Theoretical and Computational Neuroscience Oregon Health & Science University, Portland, Oregon. A unique postdoctoral position is available at the Oregon Health & Science University. Dr. Todd Leen of the Oregon Graduate Institute (OGI), and Dr. Patrick Roberts of the Neurological Sciences Institute (NSI) invite applications for a senior research associate. This NSF-funded project will investigate the sources and effects of noise in the electrosensory systems of mormyrid fish. The mormyrid is a weakly electric fish that uses is electrosensory systems for navigation, communication, and hunting. The system is experimentally well-characterized physiologically and anatomically from synaptic plasticity up through the behaving animal; much of this experimental work was and is carried out by Dr. Curtis Bell at NSI. Previous modeling work on this system is featured on Dr. Robert's page http://www.ohsu.edu/nsi/faculty/robertpa/robertpa.htm. This project will focus on stochastic dynamics of adaptation and signal processing in the mormyrid. The successful candidate will have background in statistical or theoretical physics or in neural systems modeling with facility in theory as well as computation, and a PhD in either theoretical physics, theoretical neurobiology, or an allied field. The position is open immediately. Candidates should send a CV and arrange to have three letters of recommendation sent to Todd K. Leen, Oregon Graduate Institute, 20000 NW Walker Road, Beaverton, OR 97006, or by email to tleen at cse.ogi.edu. Oregon Health & Science University is an Affirmative Action / Equal Opportunity Employer. From syounger at boulder.net Sun Sep 2 22:19:22 2001 From: syounger at boulder.net (A. Steven Younger) Date: Sun, 2 Sep 2001 19:19:22 -0700 Subject: meta-learning Message-ID: <003b01c1341e$db06be40$922b97cc@syounger> Dear Connectionists, In a connectionist message on Aug 30, 2001, Juergen Schmidhuber graciously drew attention to some work on meta-learning (learning how to learn) done by Sepp Hochreiter and myself. This work was presented by Sepp at the ICANN conference in August of this year [5]. For those of you that are interested, we have a companion paper that I presented at the IJCNN conference in July of this year [9]. This paper describes our work from a different perspective - that of "Fixed-Weight Learning." Some other related papers are: L. A. Feldkamp, et al. in 1996 [3] used a similar Fixed-Weight Learning method to meta-learn time series prediction problems. The possibility of this type of meta-learning was conjectured by N. E. Cotter and P. R. Conwell in their 1990 and 1991 [1,2] original papers on Fixed-Weight Learning. Some examples of neural networks of this type were presented by myself, Conwell and Cotter in a 1999 paper [8]. Another type of neural network that has embedded learning was presented by Schmidhuber in 1993 [6]. An explanation of the Long Short-Term Memory (LSTM) was presented by Hochreiter and Schmidhuber in 1997 [4]. A good presentation on the principles of meta-learning can be found in Schmidhueber, et al. in 1996 [7] This list is by no means exhaustive, but it should give a starting point for anyone interested in this topic. References: [1] N. E. Cotter and P. R. Conwell. "Fixed-Weight Networks Can Learn." In International Joint Conference on Neural Networks held in San Diego 1990, IEEE, New York, 1990, pp. II.553- 559 [2] N. E. Cotter and P. R. Conwell. "Learning Algorithms and Fixed Dynamics." In International Joint Conference on Neural Networks held in Seattle 1991. by IEEE. New York: IEEE 1991, pp. I.799- 804. [3] L. A. Feldkamp, G. V. Puskorius and P. C. Moore. "Adaptation from Fixed-Weight Dynamic Networks. "Proceedings of International Conference of Neural Networks 96, IEEE-1996 [4] Sepp Hochreiter and Juergen Schmidhuber, "Long Short-Term Memory." Neural Computation. 9(8) pp. 1735-1780, 1997 [5] Sepp Hochreiter, A. Steven Younger and Peter R. Conwell, "Learning to learn using gradient descent", Lecture Notes on Comp. Sci. 2130, Proc. Intl. Conf. on Artificial Neural Networks (ICANN-2001), editors G. Dorffner and H. Bischof and K. Hornik", Springer: Berlin, Heidelberg, pp. 87-94, 2001 [6] Juergen Schmidhuber. "A neural network that embeds its own meta-levels." In Proc. Of the International Conference on Neural Networks '93, San Fransisco, IEEE 1993 [7] Juergen Schmidhuber, Jieyu Zhao, and Marco Wiering. "Simple Principles of Meta-Learning", TR-IDSIA-69-96 http://www.idsia.ch [8] A. Steven Younger, P. R. Conwell, and N. E. Cotter. "Fixed-Weight On-Line Learning." IEEE Transactions on Neural Networks. Vol.10 No. 2, March 1999 pp. 272-283 [9] A. Steven Younger, Sepp Hochreiter, and Peter R. Conwell. "Meta-Learning with Backpropagation." IJCNN'01 International Joint Conference on Neural Networks 2001. IEEE-2001, pp.2001-2006 ------------------------------------------------------------------------- A. Steven Younger syounger at boulder.net Department of Computer Science Data Fusion Corporation University of Colorado at Boulder Northglenn CO Boulder CO From M.Garagnani at open.ac.uk Mon Sep 3 00:38:25 2001 From: M.Garagnani at open.ac.uk (M.Garagnani@open.ac.uk) Date: Mon, 3 Sep 2001 05:38:25 +0100 Subject: Research Fellowship in Spatial Reasoning and AI Planning Message-ID: ( Apologies if you receive multiple copies ) ( Please forward to anyone who could be interested ) ================================================== Computing Department Faculty of Mathematics and Computing The Open University, United Kingdom ---------------------------------------------------- RESEARCH FELLOW IN SPATIAL REASONING AND AI PLANNING The AI (Artificial Intelligence) group of the Computing Department at the Open University is looking for a Research Fellow to work on a new project to investigate the use of spatial reasoning and pattern-matching techniques in AI planning. The project, funded by the Engineering and Physical Sciences Research Council (EPSRC), is due to start in November 2001. The post is for 16 months. The overall aim of the project will be to demonstrate the viability of an entirely new approach to AI planning, based on multi-dimensional spatial representations of domains and problems. If successful, the results of this project are expected to mark the beginning of a new, interdisciplinary, area of research. Candidates should hold a PhD in Computing (or be close to completion) and have experience in the field of Spatial Reasoning. Knowledge of connectionist systems and/or AI planning would be an advantage. The main deliverable of the project will be a new planning system prototype which will make use of pattern-matching and spatial reasoning techniques to represent and reason about actions and states of the world. The successful candidate will take up the post as soon as possible and will be based in the Computing Department of the Faculty of Mathematics and Computing, the Open University, Milton Keynes (United Kingdom). By joining the AI Research Group, a node of the European Network of Excellence in AI Planning (see http://www.planet-noe.org/index.html), the postholder will become part of a dynamic and stimulating research team and will be able to contribute actively to the shaping of this rapidly expanding group. To learn more about the general context of the research, please look at the web site http://computing.open.ac.uk/ or telephone Prof Pat Hall on +44 (0)1825 71 2661 or +44 (0)1908 652694, email p.a.v.hall at open.ac.uk. Appointment will be made on the Research Fellow 1A salary scale in the range 17,451-21,290. For further particulars, an application pack and access details for disabled applicants, please contact the Recruitment Secretary (mcs-recruitment at open.ac.uk), or telephone +44 (0)1908 654161. Please quote ref. 7561. The closing date for applications is Friday 28th September 2001. Interviews will be on 10th October 2001. Disabled applicants whose skills and experience meet the requirements of the job will be interviewed. Please let us know if you need your copy of the further particulars in large print, on computer disk, or on audio cassette tape. Hearing impaired persons may make enquiries on Milton Keynes +44 (0)1908 654901 (Minicom answerphone). The University offers a wide range of jobs with excellent training and career development opportunities. We actively promote equal opportunities in education and employment and welcome applications from all sections of the community. =================== Dr. Max Garagnani Visiting Scholar International Computer Science Institute 1947 Center St. Berkeley, CA 94704-1198 Tel. +1 510 666 2926 Fax. +1 510 666 2956 From harnad at coglit.ecs.soton.ac.uk Mon Sep 3 06:35:46 2001 From: harnad at coglit.ecs.soton.ac.uk (Stevan Harnad) Date: Mon, 3 Sep 2001 11:35:46 +0100 (BST) Subject: Two Visual Systems: BBS Call for Commentators Message-ID: Below is the abstract of a forthcoming BBS target article Please reply to: calls at bbsonline.org Two Visual Systems and Two Theories of Perception: An Attempt to Reconcile the Constructivist and Ecological Approaches by Joel Norman http://www.bbsonline.org/Preprints/Norman/ http://psy.haifa.ac.il/~maga/tvs&ttp.pdf This article has been accepted for publication in Behavioral and Brain Sciences (BBS), an international, interdisciplinary journal providing Open Peer Commentary on important and controversial current research in the biobehavioral and cognitive sciences. Commentators must be BBS Associates or nominated by a BBS Associate. To be considered as a commentator for this article, to suggest other appropriate commentators, or for information about how to become a BBS Associate, please reply by EMAIL within three (3) weeks to: calls at bbsonline.org If you are not a BBS Associate, please approach a current BBS Associate (there are currently over 10,000 worldwide) who is familiar with your work to nominate you. All past BBS authors, referees and commentators are eligible to become BBS Associates. A full electronic list of current BBS Associates is available at this location to help you select a name: http://www.bbsonline.org/Instructions/assoclist.html If no current BBS Associate knows your work, please send us your Curriculum Vitae and BBS will circulate it to appropriate Associates to ask whether they would be prepared to nominate you. (In the meantime, your name, address and email address will be entered into our database as an unaffiliated investigator.) To help us put together a balanced list of commentators, please give some indication of the aspects of the topic on which you would bring your areas of expertise to bear if you were selected as a commentator. To help you decide whether you would be an appropriate commentator for this article, an electronic draft is retrievable from the online BBSPrints Archive, at the URL that follows the abstract below. _____________________________________________________________ Two Visual Systems and Two Theories of Perception: An Attempt to Reconcile the Constructivist and Ecological Approaches Joel Norman Department of Psychology University of Haifa Haifa, Israel jnorman at psy.haifa.ac.il KEYWORDS: Visual perception theories, ecological, constructivist, two visual systems, space perception, size perception, dual-process approach ABSTRACT: The two contrasting theoretical approaches to visual perception, the constructivist and the ecological, are briefly presented and illustrated through their analyses of space perception and size perception. Earlier calls for their reconciliation and unification are reviewed. Neurophysiological, neuropsychological, and psychophysical evidence for the existence of two quite distinct visual systems, the ventral and the dorsal, is presented. These two perceptual systems differ in their functions; the ventral system^Rs central function is that of identification, while the dorsal system is mainly engaged in the visual control of motor behavior. The strong parallels between the ecological approach and the functioning of the dorsal system and between the constructivist approach and the functioning of the ventral system are noted. It is also shown that the experimental paradigms used by the proponents of these two approaches match the functions of the respective visual systems. A dual-process approach to visual perception emerges from this analysis, with the ecological-dorsal process transpiring mainly without conscious awareness, while the constructivist-ventral process is normally conscious. Some implications of this dual-process approach to visual-perceptual phenomena are presented, with emphasis on space perception. http://www.bbsonline.org/Preprints/Norman/ http://psy.haifa.ac.il/~maga/tvs&ttp.pdf ___________________________________________________________ Please do not prepare a commentary yet. Just let us know, after having inspected it, what relevant expertise you feel you would bring to bear on what aspect of the article. We will then let you know whether it was possible to include your name on the final formal list of invitees. _______________________________________________________________________ *** SUPPLEMENTARY ANNOUNCEMENTS *** (1) The authors of scientific articles are not paid money for their refereed research papers; they give them away. What they want is to reach all interested researchers worldwide, so as to maximize the potential research impact of their findings. Subscription/Site-License/Pay-Per-View costs are accordingly access-barriers, and hence impact-barriers for this give-away research literature. There is now a way to free the entire refereed journal literature, for everyone, everywhere, immediately, by mounting interoperable university eprint archives, and self-archiving all refereed research papers in them. Please see: http://www.eprints.org http://www.openarchives.org/ http://www.cogsci.soton.ac.uk/~harnad/Tp/nature4.htm --------------------------------------------------------------------- (2) All authors in the biobehavioral and cognitive sciences are strongly encouraged to self-archive all their papers in their own institution's Eprint Archives or in CogPrints, the Eprint Archive for the biobehavioral and cognitive sciences: http://cogprints.soton.ac.uk/ It is extremely simple to self-archive and will make all of our papers available to all of us everywhere, at no cost to anyone, forever. Authors of BBS papers wishing to archive their already published BBS Target Articles should submit it to BBSPrints Archive. Information about the archiving of BBS' entire backcatalogue will be sent to you in the near future. Meantime please see: http://www.bbsonline.org/help/ and http://www.bbsonline.org/Instructions/ --------------------------------------------------------------------- (3) Call for Book Nominations for BBS Multiple Book Review In the past, Behavioral and Brain Sciences (BBS) had only been able to do 1-2 BBS multiple book treatments per year, because of our limited annual page quota. BBS's new expanded page quota will make it possible for us to increase the number of books we treat per year, so this is an excellent time for BBS Associates and biobehavioral/cognitive scientists in general to nominate books you would like to see accorded BBS multiple book review. (Authors may self-nominate, but books can only be selected on the basis of multiple nominations.) It would be very helpful if you indicated in what way a BBS Multiple Book Review of the book(s) you nominate would be useful to the field (and of course a rich list of potential reviewers would be the best evidence of its potential impact!). *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-* From glenn at bioss.sari.ac.uk Mon Sep 3 09:31:46 2001 From: glenn at bioss.sari.ac.uk (Glenn Marion) Date: Mon, 03 Sep 2001 14:31:46 +0100 Subject: JOB: Machine learning (2 posts) - Edinburgh Scotland Message-ID: <3B938642.A08E00C8@bioss.ac.uk> BIOMATHEMATICS AND STATISTICS SCOTLAND (BioSS) Permanent Position based in Edinburgh STATISTICAL AND MACHINE LEARNING TECHNIQUES FOR MOLECULAR DATA We are seeking an established research scientist (at least 5 years postgraduate experience and preferably a PhD) with strong mathematical/computational background and good communication skills. The position is permanent and has a starting salary of ?25000 - ?28800, though a higher salary may be available for an exceptional candidate. A further, fixed-term, position may be available for a new post-doctoral researcher (salary up to ?23700). Information on these and several other posts is available from http://www.bioss.sari.ac.uk/vacancies.htm An application form and further information about BioSS can be obtained from the BioSS Administrative Officer, Betty Heyburn, King's Buildings (JCMB), Edinburgh EH9 3JZ, betty at bioss.ac.uk, tel. +44 (0)131 650 4900. Informal enquiries can be made to Chris Glasbey, email chris at bioss.ac.uk, tel. +44 (0)131 650 4899. CLOSING DATE FOR APPLICATIONS: 24 SEPTEMBER 2001 -- ________________________________________________________________________ Glenn Marion Biomathematics & Statistics Scotland, The University of Edinburgh, James Clerk Maxwell Building, The King's Buildings Edinburgh EH9 3JZ glenn at bioss.ac.uk http://www.bioss.ac.uk/~glenn Tel : +44 (0)131 650 4898 Fax : +44 (0)131 650 4901 ________________________________________________________________________ From dprokhor at ford.com Mon Sep 3 11:12:59 2001 From: dprokhor at ford.com (Prokhorov, Danil (D.V.)) Date: Mon, 3 Sep 2001 11:12:59 -0400 Subject: meta-learning/fixed-weight learning without LSTM? Message-ID: <200109031513.f83FD3w25708@dymwsm09.mailwatch.com> Dear Connectionists, After two very recent messages of Juergen Schmidhuber and A. Steven Younger drawing your attention to meta-learning and fixed-weight learning, I thought I would provide you with yet another reference and the abstract of our recent paper fixed-weight learning published in Proceedings of the 11th Yale Workshop on Adaptive and Learning Systems (June 4-6, 2001). L. A. Feldkamp, D. V. Prokhorov, and T. M. Feldkamp, Conditioned Adaptive Behavior from a Fixed Neural Network, Proceedings of the Eleventh Yale Workshop on Adaptive and Learning Systems, New Haven, CT, pp. 78-83, 2001. Abstract (sorry for its conciseness) We demonstrate that a fixed-weight recurrent neural network (RMLP-style) can be trained to exhibit input-output behavior that depends on which of two conditioning tasks had been performed a substantial number of time steps in the past. This behavior can also be made to survive an intervening interference task. ____________________________________________________________________________ It may be difficult to find these Proceedings though. Please contact Danil Prokhorov (dprokhor at ford.com) or Lee Feldkamp (lfeldkam at ford.com) to receive an electronic copy of this paper. Sincerely, Danil Prokhorov Artificial Neural Systems and Fuzzy Logic Group Ford Research Laboratory 2101 Village Rd., MD 2036 Dearborn, MI 48121-2053, U.S.A. From S.M.Bohte at cwi.nl Tue Sep 4 09:12:44 2001 From: S.M.Bohte at cwi.nl (Sander Bohte) Date: Tue, 4 Sep 2001 15:12:44 +0200 Subject: preprint: paper on error-backpropagation in spiking neural networks Message-ID: Apologies if you receive this posting more than once. A preprint is available for download, of the paper 'SpikeProp: Error-Backpropagation for Networks of Spiking Neurons', by Sander Bohte, Joost Kok and Han La Poutr? (to appear in `Neurocomputing'). For a network of spiking neurons that encodes information in the timing of individual spike-times, we derive a supervised learning rule, SpikeProp, akin to traditional error-backpropagation and show how to overcome the discontinuities introduced by thresholding. Using this learning algorithm, we demonstrate how networks of spiking neurons with biologically reasonable action potentials can perform complex non-linear classification in fast temporal coding just as well as rate-coded networks. We perform experiments for the classical XOR-problem, when posed in a temporal setting, as well as for a number of other benchmark datasets. When comparing the (implicit) number of biological neurons that would be required for the respective encodings, it is empirically demonstrated that temporal coding potentially requires significantly less neurons. As we show that reliable temporal computation can only be accomplished by spike-response functions with a time constant longer than the coding interval, the results also refute the long-standing argument against temporal coding that states that the typical integration times of real neurons are too long to compute a fast temporal code. keywords: spiking neural networks, temporal coding, error-backpropagation, XOR. Download Instructions: Go to http://www.cwi.nl/~sbohte/pub_spikeprop.htm and click on the file to download (in PDF format [393K], or zipped Postscript [515K]). Comments Welcome If you have problems downloading, please e-mail me. Sander Bohte From samengo at cab.cnea.gov.ar Tue Sep 4 15:16:28 2001 From: samengo at cab.cnea.gov.ar (Ines Samengo) Date: Tue, 04 Sep 2001 16:16:28 -0300 Subject: two papers: Information maps, and Information loss in decoding Message-ID: <3B95288C.B221002D@cab.cnea.gov.ar> Dear connectionists, The following papers may be of interest to you: Measuring information spatial densities Michele Bezzi, Ines Samengo, Stefan Leutgeb and Sheri J. Mizumori Accepted in Neural Computation, 2001. Abstract: A novel definition of the stimulus-specific information is presented, which is particularly useful when the stimuli constitute a continuous and metric set, as for example, position in space. The approach allows one to build the spatial information distribution of a given neural response. The method is applied to the investigation of putative differences in the coding of position in hippocampus and lateral septum. http://www.cab.cnea.gov.ar/users/samengo/pub.html ---------------- The information loss in an optimal maximum likelihood decoding Ines Samengo Accepted in Neural Computation, 2001. Abstract: The mutual information between a set of stimuli and the elicited neural responses is compared to the corresponding decoded information. The decoding procedure is presented as an artificial flattening of the joint probabilities between stimuli and responses. The information loss is quantified. Whenever the conditional probabilities are only slightly distorted, the information loss is shown to be quadratic in the distortion. http://www.cab.cnea.gov.ar/users/samengo/pub.html _______________________________________________________________ Ines Samengo samengo at cab.cnea.gov.ar fax: +54 2944 445299 Centro Atomico Bariloche (8.400) San Carlos de Bariloche Rio Negro Argentina _______________________________________________________________ From matsudaira at gmd.gr.jp Tue Sep 4 22:47:06 2001 From: matsudaira at gmd.gr.jp (Yukako Matsudaira) Date: Wed, 5 Sep 2001 11:47:06 +0900 Subject: Researchers Post Offerings for GMD Japan Research Laboratory Message-ID: <006501c135b5$0d2d2660$0202630a@yukakopc> Research Positions - Announcement ---------------------------------------------------------------------- Research Positions in Autonomous Robotics in Open and Dynamic Environments ------------------------------------------------------------------------ Seven new research positions (including post-doctoral positions) are open at GMD-Japan Research Laboratory, Kitakyushu, Japan and will be filled at the earliest convenience. The laboratory (planned to be a team of eight scientists and two support staff) is based on long-term cooperations with the Japanese research community and the (local) companies and organizations. It focuses on robotics in two main research fields, namely autonomous inspection of sewer pipes and soccer-playing robot teams, as known from the RoboCup competitions. Investigated questions are: - How to self-localize and move in many DoF without global correlation? - Interpretation / integration / abstraction / compression of complex sensor signals? - How to build models based upon sensor-motor coordination for (partially) unknown environments? - How to predict sensor-actuator correlations and generate motor signals appropriately? - How to coordinate multiple loosely coupled robots? Autonomous mobile robotics for sewer inspection and soccer-playing are regarded as one of the most promising experimental and application environment in this context. Real dynamic environments and real autonomy (which is required in these setups), settle these questions in concrete environments. The overall goal is of course not only implementing prototype systems, but to get a better understanding of autonomy, and situatedness. Modeling, adaptation, clustering, prediction, communication, or - from the perspective of robotics - spatial and behavioral modeling, localization, navigation, and exploration are cross-topics addressed in most questions. It is also very important to connect this with possible applications to be explored with (local) companies, like sewer operational units or edutainment robotics manufacturers. Techniques to be employed and developed include dynamical systems, connectionist approaches, behavior-based techniques, rule based systems, and systems theory. Experiments are based on physical robots. Thus the discussion of experimental setups and particularly the meaning of embodiment and situatedness are natural topics. GMD-JRL is an independent research unit embedded and located at the newly opened Science and Research Park (run by the Kitakyushu Foundation for the Advancement of Industry, Science and Technology FAIS http://www.city.kitakyushu.jp). Kitakyushu is located at the coast line of the Sea of Japan on Kyushu Island. It is as far away from Tokyo as from Shanghai and a traditional harbour gate to south-east Asia. In addition, there are close connections to the Fraunhofer Institute of Autonomous Intelligent Systems (FhI AIS, http://www.fraunhofer.ais.gmd.de/ ) headed by Prof. Dr. Thomas Christaller in Sankt Augustin, Germany. The technical environment is up-to-date and the laboratory itself is located in four modern office rooms on the campus of the Science and Research Park, which can be reached by a 30 minutes ride from central Kitakyushu. If the above challenges arise your interest, please proceed to our expectations regarding the ideal candidate: - Master/Diploma or Ph.D. / doctoral degree in computer sciences, electrical or mechanical engineering, physics, mathematics, biology, or related disciplines: - Experiences in experimenting with autonomous systems. - Theoretical foundations in mathematics, control theory, connectionism, dynamical systems, or systems theory. - Interest in joining an international team of highly motivated researchers. Furthermore it is expected that the candidate engages her/himself to bring in her/his own perspective to blend it with the whole group. There are many opportunities for writing a Ph.D. thesis. Salary starts at 7.8 Mill. Yen per year depending on experience and graduation. For any further information, and applications (including addresses of referees, two recent publications, and a letter of interest) please contact: Prof. Dr. Thomas Christaller, Director of GMD-Japan Research Laboratory Secretary: Yukako Matsudaira mailto:matsudaira at gmd.gr.jp Tel: +81-93-695-3085 Fax: +81-93-695-3525 URL: http://www.gmd.de/JRL/ -- _______________________________________________________ Visit the AiS Colloquium on Autonomy every two weeks on Wednesday: http://www.ais.fraunhofer.de/kolloquium/index.html Visit the European Conference on Artificial Life at Prague http://www.cs.cas.cz/ecal2001 Visit the web portal on Edutainment Robotics http://www.edutainment-robotics.org Visit the RoboFesta in Japan http://www.robofesta.net Notice the Interdisciplinary College at G=FCnne http://www.tzi.de/ik2002 _______________________________________________________ Prof. Dr. rer.nat. Thomas Christaller mailto:thomas.christaller at ais.fraunhofer.de http://www.ais.fraunhofer.de/people/Thomas.Christaller/ Fraunhofer Institute for Autonomous intelligent Systems - AiS http://www.ais.fraunhofer.de/ Schloss Birlinghoven D-53754 Sankt Augustin Phone: +49-2241-14-2678, Fax +49-2241-14-2384 Secretary: mailto:Renate.Henkeler at ais.fraunhofer.de, mailto:Eva.Sommer at ais.fraunhofer.de _______________________________________________________ From aapo at james.hut.fi Wed Sep 5 06:37:38 2001 From: aapo at james.hut.fi (Aapo Hyvarinen) Date: Wed, 5 Sep 2001 13:37:38 +0300 Subject: New Book on ICA Message-ID: [Our apologies if you receive multiple copies of this message.] The following book is now available: INDEPENDENT COMPONENT ANALYSIS Aapo Hyvarinen, Juha Karhunen, and Erkki Oja Published by John Wiley & Sons, 2001. 504 pages. Independent Component Analysis (ICA) is one of the most exciting topics in the fields of neural computation, advanced statistics, and signal processing. This is the first book to provide a comprehensive introduction to this new technique complete with the mathematical background needed to understand and utilize it. It offers a general overview of the basics of ICA, important solutions and algorithms, and in-depth coverage of new applications in vision research, brain imaging, telecommunications, and more. See the book web page at http://www.cis.hut.fi/projects/ica/book/ ---------------------------------------------------- Aapo Hyvarinen Neural Networks Research Centre Helsinki University of Technology P.O.Box 5400, FIN-02015 HUT, Finland Tel: +358-9-4513278, Fax: +358-9-4513277 Email: Aapo.Hyvarinen at hut.fi Home page: http://www.cis.hut.fi/aapo/ ---------------------------------------------------- From jianfeng at cogs.susx.ac.uk Wed Sep 5 09:53:32 2001 From: jianfeng at cogs.susx.ac.uk (Jianfeng Feng) Date: Wed, 5 Sep 2001 14:53:32 +0100 (BST) Subject: Ph.D studentship Message-ID: We are interested in applicants having backgrounds in any of the following Statistics Mathematics Computer Science Physics Sussex has an excellent research reputation (12th in the latest league table). The school of cognitive and computing sciences at the university of Sussex grew out of a pioneering multi-disciplinary centre for research into intelligent systems and the mechanisms underlying them. Artificial intelligence and Neuroscience are areas in which Sussex is exceptionally strong, a view supported by the results of the last three national Research Assessment Exercises in which the computer science and artificial intelligence group in the school received 5 rating. Applications are invited for one research studentship to be held from 1st January 2002 for three years. The studentship is funded by EPSRC. The student will work on learning in spiking neuronal networks. Applicants should have a 1st or 2(i) class honours degree in an appropriate subject. Further particulars on my interests etc. could be found at http://www.cogs.susx.ac.uk/users/jianfeng Applications should be sent as soon as possible to: Dr. J. Feng, Cogs, Sussex University, BN1 9QH, UK, e-mail: jianfeng at cogs.susx.ac.uk, or jf218 at cam.ac.uk, tel. 0044 1273 678062. From smyth at ics.uci.edu Thu Sep 6 13:35:22 2001 From: smyth at ics.uci.edu (Padhraic Smyth) Date: Thu, 06 Sep 2001 10:35:22 -0700 Subject: faculty positions in new Statistics Dept at UCI Message-ID: <3B97B3DA.CBDD94B0@ics.uci.edu> [apologies as usual if you receive this more than once] Dear Colleagues, The University of California, Irvine (UCI) has just announced the creation of a new statistics department. In the coming year UCI will be hiring for three positions, two tenured (one of which is the chair) and one tenure-track. Computational statistics (in a broad sense, including graphical models, machine learning, etc) is one of the areas where we are likely to be hiring, so I would encourage those of you who are interested to please apply and/or to pass the word along to any colleagues who might be interested. (full details below). This new department is likely to offer significant opportunities for interdisciplinary collaborations with faculty and students in other departments at UCI including computer science, biological sciences, medicine, cognitive science, economics, all of which are actively involved in setting up the new department. Padhraic Smyth Information and Computer Science University of California, Irvine smyth at ics.uci.edu UNIVERSITY OF CALIFORNIA, IRVINE Announces a Department in Statistics The University of California, Irvine is starting a Department of Statisitcs. We anticipate appointing a full-time faculty in statistics of 6-8 people over the next seven years, with several more half-time appointments shared with other units at UCI. It will be a department with a strongly interdisciplinary flavor, focused both on theoretical research and applied problems, and it will be associated with an independent statistical consulting center that presently serves the campus and surrounding businesses and industry. Three faculty positions are open for recruitment in 2001-02: two with tenure, including one for the Chair of the Department; and one tenure-track assistant professorship. In all cases demonstrated excellence in research, teaching, and service are sought and, in the case of the Chair evidence of leadership ability is required. UCI is one of the youngest campuses in the University of California, yet we are already ranked 10th among public universities. We are projected to grow by almost 50% over the next ten years, with significant increases in graduate enrollment. The new Department of Statistics will occupy a prominent place in this expanding academic profile, and it will interact significantly with a wide range of existing departments and innovative programs across the whole campus. A detailed account of plans for the new department is available at www.evc.uci.edu/proposal.pdf. For information about UCI, see www.uci.edu/. For information about the community around UCI, see www.oc.ca.gov/. Completed applications with a cover letter, sample research publications, and if possible three letters of recommendation and up to five additional names who may be contacted should be sent to the Chair of the search committee: Professor Duncan Luce c/o Office of the Executive Vice Chancellor 535 Administration Bldg. University of California, Irvine Irvine, CA 92697-1000 (Applications received before November 1, 2001, will receive preference, but later applications will be considered.) The University of California is an Equal Opportunity Employer committed to excellence through diversity. From P.J.Lisboa at livjm.ac.uk Fri Sep 7 04:45:28 2001 From: P.J.Lisboa at livjm.ac.uk (Lisboa Paulo) Date: Fri, 7 Sep 2001 09:45:28 +0100 Subject: Workshop on Regulatory Issues in Medical Decision Support, UCL, F riday 19th October 2001 Message-ID: Dear Colleagues, Europe and the USA have rapidly growing markets for medical equipment, whose operation increasingly relies on complex software. The explosion of networking and databases, coupled with user demands for better quality, personalisation and remote care, all point to a fast commercial development of healthcare informatics. However, there are many unresolved issues surrounding the application of the Medical Devices Directives, especially in Medical Decision Support. This workshop brings together the main complementary perspectives that impact directly on regulatory procedures, introduced by speakers who are recognised authorities in the design, evaluation and practical use of statistical, rule-based and neural network systems. The purpose of the workshop is to identify and discuss key issues, so as to provide pointers to effective and practical ways to resolve them. A programme and registration leaflet is attached. In time a link will appear also at the AIME website http://www.aime.org.uk/ Paulo Lisboa. ____________________________________________________________________________ PROGRAMME 09.30-10.15 Registration and Coffee 10.25-10.30 Welcome Professor Paulo Lisboa, John Moores University, Liverpool. 10.30-11.10 The Medical Devices Directives: Application of the Directives to Medical Decision Support Systems (Scope and Conformity Assessment) Mr. P. Stonebrook, Medical Devices Agency. 11.10-11.50 Key issues for computer-based decision support in clinical practice Dr. Bipin Vadher, Bromley Hospitals NHS Trust. 11.50-12.30 Experience with longstanding decision support and perspectives of new developments in Europe. Dr. Susan Clamp, Clinical Information Science Unit, Leeds. 12.30-13.30 Lunch 13.30-14.15 Issues in the development of good statistical models for prognosis Professor Doug Altman, Centre for Statistics in Medicine, Oxford. 14.15-15.00 Clinical Decision- making by machine: maximising safety and limiting risk Professor John Fox, ICRF, Lincoln's Inn Fields. 15.00-15.15 Tea 15.15-16.00 Validation of neural network medical systems Dr. Ian Nabney, Cardionetics Institute of Bioinformatics, Aston University. 16.00-16.30 Discussion Dr. Jeremy Wyatt, UCL, to lead on practical ways forward in the evaluation and certification of software for medical decision support. 16.30 Close of Meeting ____________________________________________________________________________ _______ REGISTRATION LEAFLET Regulatory Issues in Medical Decision Support 19th October, University College London REGISTRATION FEES:(Incl. of lunch and coffee) Members of AIME Institutions and BMIS ? 60.00 Non-Members ? 75.00 Student / Retired ? 40.00 Special requirements (e.g. dietary, disabled) Title: Name: Address: Post Code: Tel: Fax: Email: METHOD OF PAYMENT (Cheques payable to 'The Institute of Physics & Engineering in Medicine') I wish to attend the above meeting and enclose my registration fee. I wish to attend the above meeting. Please debit my credit card as detailed below. (Barclaycard / Access / Visa / Mastercard / Eurocard) Expiry Date / Signature I wish to attend the above meeting. Please send an invoice for my registration fee to the address given below. My Purchase Order Number is _________________ Please note that invoices cannot be supplied without an official purchase order. PLEASE NOTE THAT REGISTRATIONS CANNOT BE ACCEPTED WITHOUT EITHER ADVANCE PAYMENT OR AN OFFICIAL PURCHASE ORDER. Please return completed form to:IPEM Meetings Fairmount House, 230 Tadcaster Road, York YO24 1ES Fax: 01904 612279 To be received no later than 5 October 2001. From kstanley at cs.utexas.edu Fri Sep 7 17:08:06 2001 From: kstanley at cs.utexas.edu (Kenneth Owen Stanley) Date: Fri, 7 Sep 2001 16:08:06 -0500 (CDT) Subject: Neuroevolution paper, software, and demo Message-ID: <200109072108.QAA22092@wheat.cs.utexas.edu> Dear Connectionists, NeuroEvolution of Augmenting Topologies (NEAT) is a new approach to evolving the topology and weights of artificial neural networks. Neural network topologies in NEAT start minimally and grow increasingly complex over generations. A paper describing the method (abstract below), source code, and animated demos in a robot control domain are all available at: http://www.cs.utexas.edu/users/nn/pages/research/ne-methods.html#NEAT --Kenneth Stanley and Risto Miikkulainen Paper: ----------------------------------------------------------------------- EVOLVING NEURAL NETWORKS THROUGH AUGMENTING TOPOLOGIES Kenneth O. Stanley and Risto Miikkulainen Department of Computer Sciences, The University of Texas at Austin Technical Report TR-AI-01-290, June 2001. http://www.cs.utexas.edu/users/nn/pages/publications/abstracts.html#stanley.utcstr01.ps.gz An important question in neuroevolution is how to gain an advantage from evolving neural network topologies along with weights. We present a method, NeuroEvolution of Augmenting Topologies (NEAT) that outperforms the best fixed-topology method on a challenging benchmark reinforcement learning task. We claim that the increased efficiency is due to (1) employing a principled method of crossover of different topologies, (2) protecting structural innovation using speciation, and (3) incrementally growing from minimal structure. We test this claim through a series of ablation studies that demonstrate that each component is necessary to the system as a whole and to each other. What results is significantly faster learning. NEAT is also an important contribution to GAs because it shows how it is possible for evolution to both optimize and complexify solutions simultaneously, offering the possibility of evolving increasingly complex solutions over generations, and strengthening the analogy with biological evolution. Software: ----------------------------------------------------------------------- NEAT C++ SOURCE CODE http://www.cs.utexas.edu/users/nn/pages/software/abstracts.html#neat-cpp Kenneth Stanley The NEAT package contains source code implementing the NeuroEvolution of Augmenting Topologies method. The source code is written in C++. The package includes implementations of experiments for XOR, single pole balancing, and both Markovian and non-Markovian double pole balancing. A 17-page postscript documentation file is included to make getting started easier. Demo: ----------------------------------------------------------------------- COMPETITIVE COEVOLUTION ROBOT CONTROL DEMO http://www.cs.utexas.edu/users/nn/pages/research/neatdemo.html Kenneth Stanley The page contains links to movies (in the form of GIF animations) depicting evolved robot controllers controlling simulated Khepera-like robots in a real-time pursuit and evasion task. The controllers were evolved using the NEAT method. The movies show the robots' behavior in the task, and also the actual evolved neural network controllers with neurons firing in real-time, so you can see what the robots are "thinking" as they compete. The main point of the demo is to demonstrate that NEAT establishes an evolutionary "arms race", i.e. is able to discover increasingly complex behaviors in competitive coevolution. From Steve.Minton at fetch.com Sun Sep 9 17:20:44 2001 From: Steve.Minton at fetch.com (Steve Minton) Date: Sun, 9 Sep 2001 14:20:44 -0700 Subject: New JAIR paper Message-ID: It was suggested to me that members of this mailing list may be interested in the following announcement: Bhattacharyya, C. and Keerthi, S.S. (2001) "Mean Field Methods for a Special Class of Belief Networks", Volume 15, pages 91-114. Available in PDF, PostScript and compressed PostScript. For quick access via your WWW browser, use this URL: http://www.jair.org/abstracts/bhattacharyya01a.html More detailed instructions are below. Abstract: The chief aim of this paper is to propose mean-field approximations for a broad class of Belief networks, of which sigmoid and noisy-or networks can be seen as special cases. The approximations are based on a powerful mean-field theory suggested by Plefka. We show that Saul, Jaakkola and Jordan' s approach is the first order approximation in Plefka's approach, via a variational derivation. The application of Plefka's theory to belief networks is not computationally tractable. To tackle this problem we propose new approximations based on Taylor series. Small scale experiments show that the proposed schemes are attractive. The article is available via: -- comp.ai.jair.papers (also see comp.ai.jair.announce) -- World Wide Web: The URL for our World Wide Web server is http://www.jair.org/ For direct access to this article and related files try: http://www.jair.org/abstracts/bhattacharyya01a.html -- Anonymous FTP from either of the two sites below. Carnegie-Mellon University (USA): ftp://ftp.cs.cmu.edu/project/jair/volume15/bhattacharyya01a.ps The University of Genoa (Italy): ftp://ftp.mrg.dist.unige.it/pub/jair/pub/volume15/bhattacharyya01a.ps The compressed PostScript file is named bhattacharyya01a.ps.Z (159K) For more information about JAIR, visit our WWW or FTP sites, or contact jair-ed at isi.edu From rampon at tin.it Sun Sep 9 18:02:29 2001 From: rampon at tin.it (Salvatore Rampone) Date: Mon, 10 Sep 2001 00:02:29 +0200 Subject: New genetic database and paper available Message-ID: Dear Connectionists, This new genetic dataset may be of interest to you: Database Name: HS3D - Homo Sapiens Splice Site Dataset URL: http://www.sci.unisannio.it/docenti/rampone/ The data base is described in the paper HS3D - Homo Sapiens Splice Site Dataset by Pollastro, P., Rampone, S. Universit del Sannio - ITALY Accepted in Nucleic Acids Research 2002 Database Issue The extended abstract is available at http://space.tin.it/scienza/srampone/ (publication page) or directly from http://space.tin.it/scienza/srampone/ramp0201.pdf. ------- Abstract: In the last years many computational tools for gene identification and characterization, mostly based on machine learning approaches, have been used. In the machine learning approach, a learning algorithm receives a set of training examples, each labelled as belonging to a particular class. The algorithm's goal is to produce a classification rule for correctly assigning new examples to these classes. The success of these methods depends largely on the quality of the data sets that are used as the training set. Furthermore a common data set is necessary when the prediction accuracy of different programs needs to be comparatively assessed. The Irvine Primate Splice Junctions Dataset (UCI Machine Learning Repository http://www.ics.uci.edu/~mlearn/MLRepository.html) is a standard "de facto" in the machine learning community, but it is now very out of date and does not include sufficient material for the most learning algorithm needs. A recent and EST confirmed data set has the same limitation in the data extend. More recently Burset et al. developed an extensive data base, but the data do not include false splice sites (negative examples), and, specifically, proximal false splice sites. The latter data form a well known critical point of classification systems. We developed a new database (HS3D - Homo Sapiens Splice Site Dataset) of Homo Sapiens Exon, Intron and Splice regions. The aim of this data set is to give standardized material to train and to assess the prediction accuracy of computational approaches for gene identification and characterization. From pam_reinagel at hms.harvard.edu Mon Sep 10 11:49:12 2001 From: pam_reinagel at hms.harvard.edu (Pamela Reinagel) Date: Mon, 10 Sep 2001 11:49:12 -0400 Subject: conference: natural stimulus statistics Message-ID: We are pleased to announce a new Gordon Research Conference Sensory coding and the natural environment: Probabilistic models of perception June 30 - July 5, 2002 Mount Holyoke College, MA. This conference will bring together participants from many disciplines to discuss the statistical structure of natural sensory stimuli, and how biological systems may use these statistics to process natural signals. The meeting grew out of a smaller workshop and will cover similar topics: From 1997: www.klab.caltech.edu/~pam/nssmeeting From 2000: www.klab.caltech.edu/~pam/nss2000.html As the date approaches, details and applications will be available on the GRC website (http://www.grc.uri.edu). Until then, please mark the date on your calendars! Pam Reinagel and Bruno Olshausen co-chairs, 2002 GRC From rampon at tin.it Mon Sep 10 14:01:47 2001 From: rampon at tin.it (Salvatore Rampone) Date: Mon, 10 Sep 2001 20:01:47 +0200 Subject: Paper address correction - New genetic database and paper available Message-ID: Dear Connectionists, with reference to the message "New genetic database and paper available" posted to this list, the paper downlod address is http://space.tin.it/scienza/srampone/ramp0201.zip and not http://space.tin.it/scienza/srampone/ramp0201.pdf. Sorry for the mistake. -salvatore ----------- Salvatore Rampone Facolta' di Scienze MM.FF.NN. and INFM Universita' del Sannio Via Port'Arsa 11 I-82100 Benevento ITALY E-mail: rampone at unisannio.it From: esann To: "Connectionists at cs.cmu.edu" References: From bogus@does.not.exist.com Thu Sep 13 10:49:38 2001 From: bogus@does.not.exist.com () Date: Thu, 13 Sep 2001 16:49:38 +0200 Subject: CFP: ESANN'2002 European Symposium on Artificial Neural Networks Message-ID: ---------------------------------------------------- | | | ESANN'2002 | | | | 10th European Symposium | | on Artificial Neural Networks | | | | Bruges (Belgium) - April 24-25-26, 2002 | | | | Announcement and call for papers | ---------------------------------------------------- Technically co-sponsored by the IEEE Region 8*, the IEEE Benelux Section, the International Neural Networks Society and the European Neural Networks Society (* to be confirmed). The call for papers for the ESANN'2002 conference is now available on the Web: http://www.dice.ucl.ac.be/esann For those of you who maintain WWW pages including lists of related ANN sites: we would appreciate if you could add the above URL to your list; thank you very much! We try as much as possible to avoid multiple sendings of this call for papers; however please apologize if you receive this e-mail twice, despite our precautions. You will find below a short version of this call for papers, without the instructions to authors (available on the Web). If you have difficulties to connect to the Web please send an e-mail to esann at dice.ucl.ac.be and we will send you a full version of the call for papers. ESANN'2002 is organised in collaboration with the UCL (Universite catholique de Louvain, Louvain-la-Neuve) and the KULeuven (Katholiek Universiteit Leuven). Scope and topics ---------------- Since its first happening in 1993, the European Symposium on Artificial Neural Networks has become the reference for researchers on fundamentals and theoretical aspects of artificial neural networks. Each year, around 100 specialists attend ESANN, in order to present their latest results and comprehensive surveys, and to discuss the future developments in this field. The ESANN'2002 conference will focus on fundamental aspects of ANNs: theory, models, learning algorithms, mathematical and statistical aspects, in the context of function approximation, classification, control, time-series prediction, signal processing, vision, etc. Papers on links and comparisons between ANNs and other domains of research (such as statistics, data analysis, signal processing, biology, psychology, evolutive learning, bio-inspired systems, etc.) are encouraged. Papers will be presented orally (no parallel sessions) and in poster sessions; all posters will be complemented by a short oral presentation during a plenary session. It is important to mention that it is the topics of the paper which will decide if it better fits into an oral or a poster session, not its quality. The selection of posters will be identical to oral presentations, and both will be printed in the same way in the proceedings. Nevertheless, authors have the choice to indicate on the author submission form that they only accept to present their paper orally. The following is a non-exhaustive list of topics covered during the ESANN conferences: - Models and architectures - Learning algorithms - Theory - Mathematics - Statistical data analysis - Classification - Approximation of functions - Time series forecasting - Nonlinear dimension reduction - Multi-layer Perceptrons - RBF networks - Self-organizing maps - Vector quantization - Support Vector Machines - Recurrent networks - Fuzzy neural nets - Hybrid networks - Bayesian neural nets - Cellular neural networks - Signal processing - Independent component analysis - Natural and artificial vision - Adaptive control - Identification of non-linear dynamical systems - Biologically plausible networks - Bio-inspired systems - Cognitive psychology - Evolutiv learning - Adaptive behaviour Special sessions ---------------- Special sessions will be organized by renowned scientists in their respective fields. Papers submitted to these sessions are reviewed according to the same rules as any other submission. Authors who submit papers to one of these sessions are invited to mention it on the author submission form; nevertheless, submissions to the special sessions must follow the same format, instructions and deadlines as any other submission, and must be sent to the same address. Here is the list of special sessions that will be organized during the ESANN'2002 conference: 1. Perspectives on Learning with Recurrent Networks (B. Hammer, J.J. Steil) 2. Representation of high-dimensional data (A. Gurin-Dugu, J. Hrault) 3. Neural Network Techniques in Fault Detection and Isolation (S. Simani) 4. Hardware and Parallel Computer Implementations of Neural Networks (U. Seiffert) 5. Exploratory Data Analysis in Medicine and Bioinformatics (A. Wismller, T. Villmann) 6. Neural Networks and Cognitive Science (H. Paugam-Moisy, D. Puzenat) A short description of these sessions will be inserted on the ESANN Web site in the next few days, and tentatively sent to this distribution list. Location -------- The conference will be held in Bruges (also called "Venice of the North"), one of the most beautiful medieval towns in Europe. Bruges can be reached by train from Brussels in less than one hour (frequent trains). The town of Bruges is world-wide known, and famous for its architectural style, its canals, and its pleasant atmosphere. The conference will be organised in a hotel located near the centre (walking distance) of the town. There is no obligation for the participants to stay in this hotel. Hotels of all level of comfort and price are available in Bruges; there is a possibility to book a room in the hotel of the conference at a preferential rate through the conference secretariat. A list of other smaller hotels is also available. The conference will be held at the Novotel hotel, Katelijnestraat 65B, 8000 Brugge, Belgium. Call for contributions ---------------------- Prospective authors are invited to submit six copies of their manuscript (including at least two originals or very good copies without glued material, which will be used for the proceedings) one signed copy of the author submission form a floppy disk or a CD (PC format preferred) containing their contribution in (generic) PostScript format before December 3, 2001. Sorry, electronic or fax submissions are not accepted. The working language of the conference (including proceedings) is English. The instructions to authors, together with the author submission form, are available on the ESANN Web server: http://www.dice.ucl.ac.be/esann Authors must indicate their choice for oral or poster presentation on the author submission form. They must also sign a written agreement that they will register to the conference and present the paper in case of acceptation of their submission. Authors of accepted papers will have to register before February 28, 2002. They will benefit from the advance registration fee. Submissions must be sent to: Michel Verleysen UCL - DICE 3, place du Levant B-1348 Louvain-la-Neuve Belgium esann at dice.ucl.ac.be All submissions will be acknowledged by fax or email before the end of December, 2001. Deadlines --------- Submission of papers December 3, 2001 Notification of acceptance February 3, 2002 Symposium April 24-26, 2002 Registration fees ----------------- registration before registration after March 15, 2002 March 15, 2002 Universities 420 450 Industries 520 550 The registration fee includes the attendance to all sessions, the ESANN'2002 dinner, a copy of the proceedings, daily lunches (24-26 April 2002), and the coffee breaks. Conference secretariat ---------------------- ESANN'2002 d-side conference services phone: + 32 2 730 06 11 24 av. L. Mommaerts Fax: + 32 2 730 06 00 B - 1140 Evere (Belgium) E-mail: esann at dice.ucl.ac.be http://www.dice.ucl.ac.be/esann Steering and local committee (to be confirmed) ---------------------------- Hugues Bersini Univ. Libre Bruxelles (B) Franois Blayo Prfigure (F) Marie Cottrell Univ. Paris I (F) Jeanny Hrault INPG Grenoble (F) Bernard Manderick Vrije Univ. Brussel (B) Eric Noldus Univ. Gent (B) Jean-Pierre Peters FUNDP Namur (B) Joos Vandewalle KUL Leuven (B) Michel Verleysen UCL Louvain-la-Neuve (B) Scientific committee (to be confirmed) -------------------- Edoardo Amaldi Politecnico di Milano (I) Herv Bourlard IDIAP Martigny (CH) Joan Cabestany Univ. Polit. de Catalunya (E) Colin Campbell Bristol Univ. (UK) Stphane Canu Inst. Nat. Sciences App. (F) Holk Cruse Universitt Bielefeld (D) Eric de Bodt Univ. Lille II & UCL Louv.-la-N. (B) Dante Del Corso Politecnico di Torino (I) Wlodek Duch Nicholas Copernicus Univ. (PL) Marc Duranton Philips / LEP (F) Richard Duro Univ. Coruna (E) Jean-Claude Fort Universit Nancy I (F) Colin Fyfe Univ. Paisley (UK) Stan Gielen Univ. of Nijmegen (NL) Marco Gori Univ. Siena (I) Bernard Gosselin Fac. Polytech. Mons (B) Manuel Grana UPV San Sebastian (E) Anne Gurin-Dugu INPG Grenoble (F) Barbare Hammer Univ. of Osnabruck (D) Martin Hasler EPFL Lausanne (CH) Laurent Hrault CEA-LETI Grenoble (F) Gonzalo Joya Univ. Malaga (E) Christian Jutten INPG Grenoble (F) Juha Karhunen Helsinky Univ. of Technology (FIN) Vera Kurkova Acad. of Science of the Czech Rep. (CZ) Jouko Lampinen Helsinki Univ. of Tech. (FIN) Petr Lansky Acad. of Science of the Czech Rep. (CZ) Mia Loccufier Univ. Gent (B) Erzsebet Merenyi Rice Univ. (USA) Jean Arcady Meyer Univ. Pierre et Marie Curie - Paris 6 (F) Jos Mira UNED (E) Jean-Pierre Nadal Ecole Normale Suprieure Paris (F) Gilles Pags Univ. Pierre et Marie Curie - Paris 6 (F) Thomas Parisini Politecnico di Milano (I) Hlne Paugam-Moisy Univ. Lumire Lyon 2 (F) Alberto Prieto Universitad de Granada (E) Leonardo Reyneri Politecnico di Torino (I) Tamas Roska Hungarian Academy of Science (H) Jean-Pierre Rospars INRA Versailles (F) Jose Santos Reyes Univ. Coruna (E) Jochen Steil Univ. Bielefeld (D) John Stonham Brunel University (UK) Johan Suykens KUL Leuven (B) John Taylor Kings College London (UK) Claude Touzet IUSPIM Marseilles (F) Marc Van Hulle KUL Leuven (B) Thomas Villmann Univ. Leipzig (D) Christian Wellekens Eurecom Sophia-Antipolis (F) ======================================================== ESANN - European Symposium on Artificial Neural Networks http://www.dice.ucl.ac.be/esann * For submissions of papers, reviews,... Michel Verleysen Univ. Cath. de Louvain - Microelectronics Laboratory 3, pl. du Levant - B-1348 Louvain-la-Neuve - Belgium tel: +32 10 47 25 51 - fax: + 32 10 47 25 98 mailto:esann at dice.ucl.ac.be * Conference secretariat d-side conference services 24 av. L. Mommaerts - B-1140 Evere - Belgium tel: + 32 2 730 06 11 - fax: + 32 2 730 06 00 mailto:esann at dice.ucl.ac.be ======================================================== From felix at idsia.ch Fri Sep 14 10:40:09 2001 From: felix at idsia.ch (Felix Gers) Date: Fri, 14 Sep 2001 16:40:09 +0200 (CEST) Subject: LSTM recurrent nets, PhD Thesis, Papers & Code Message-ID: Dear Connectionists, I am glad to announce my PhD thesis on Long Short-Term Memory (LSTM) in Recurrent Neural Networks (RNNs), several LSTM papers, and LSTM source code. Felix Gers, IDSIA www.idsia.ch -------------------------------PHD THESIS------------------------------ Long Short-Term Memory in Recurrent Neural Networks: http://www.idsia.ch/~felix/My_papers/phd.ps.gz http://www.idsia.ch/~felix/My_papers/phd.pdf On-line abstract: http://www.idsia.ch/~felix/My_papers/phd/node3.html -------------------------------JOURNAL PAPERS-------------------------- F. A. Gers, J. Schmidhuber, and F. Cummins. Learning to forget: Continual Prediction with LSTM. Neural Computation, 2000. http://www.idsia.ch/~felix/My_papers/FgGates-NC.ps.gz http://www.idsia.ch/~felix/My_papers/FgGates-NC.pdf Abstract. Long Short-Term Memory (LSTM, Hochreiter & Schmidhuber, 1997) can solve numerous tasks not solvable by previous learning algorithms for recurrent neural networks (RNNs). We identify a weakness of LSTM networks processing continual input streams that are not a priori segmented into subsequences with explicitly marked ends at which the network's internal state could be reset. Without resets, the state may grow indefinitely and eventually cause the network to break down. Our remedy is a novel, adaptive "forget gate" that enables an LSTM cell to learn to reset itself at appropriate times, thus releasing internal resources. We review illustrative benchmark problems on which standard LSTM outperforms other RNN algorithms. All algorithms (including LSTM) fail to solve continual versions of these problems. LSTM with forget gates, however, easily solves them in an elegant way. --- F. A. Gers and J. Schmidhuber. LSTM recurrent networks learn simple context free and context sensitive languages. IEEE Transactions on Neural Networks, 2001. http://www.idsia.ch/~felix/My_papers/L-IEEE.ps.gz http://www.idsia.ch/~felix/My_papers/L-IEEE.pdf Abstract. Previous work on learning regular languages from exemplary training sequences showed that Long Short-Term Memory (LSTM) outperforms traditional recurrent neural networks (RNNs). Here we demonstrate LSTM's superior performance on context free language (CFL) benchmarks for recurrent neural networks (RNNs), and show that it works even better than previous hardwired or highly specialized architectures. To the best of our knowledge, LSTM variants are also the first RNNs to learn a simple context sensitive language (CSL), namely a^n b^n c^n. -------------------------------OTHER PAPERS---------------------------- Numerous additional LSTM conference papers and TRs available at: http://www.idsia.ch/~felix/Publications.html -------------------------------LSTM CODE------------------------------- C++ and Matlab code of the LSTM algorithm available at: http://www.idsia.ch/~felix/SourceCode_Data.html -------------------------------PhD POSITION---------------------------- New LSTM PhD position at IDSIA: http://www.idsia.ch/~juergen/phd2001.html ----------------------------------------------------------------------- From maass at igi.tu-graz.ac.at Fri Sep 14 12:59:49 2001 From: maass at igi.tu-graz.ac.at (Wolfgang Maass) Date: Fri, 14 Sep 2001 18:59:49 +0200 Subject: computing on spike trains Message-ID: <3BA23785.8B332C66@igi.tu-graz.ac.at> A preprint of the following paper is now online available: REAL-TIME COMPUTING WITHOUT STABLE STATES: A NEW FRAMEWORK FOR NEURAL COMPUTATION BASED ON PERTURBATIONS by Wolfgang Maass, Thomas Natschlger, and Henry Markram. (Graz Univ. of Technology, Austria, and Weizmann Institute, Israel) ABSTRACT: This paper has the goal to establish a theoretical framework for computations on spike trains that can also be applied to biologically realistic models for recurrent circuits of spiking neurons. This new theoretical framework, the liquid state machine, differs strongly from the computational models that have emerged from computer science and artificial neural networks: it is not based on transitions between stable internal states or attractors, but rather exploits the natural transient dynamics of recurrent neural circuits as a potentially powerful analog memory device. It directs attention to the investigation of trajectories of transient internal states in very high dimensional dynamical systems, thereby providing a complement to the analysis of attractors in low dimensional dynamical systems that have so far been used as primary sources of inspiration for understanding the dynamics of neural computation. Like the Turing machine this model allows for basically unlimited computational power under idealized conditions, but for real-time computing on time-varying inputs with fading memory (rather than for offline-computing on static discrete inputs like the Turing machine). Based on this new framework we have for the first time been able to carry out complex real-time computations on spike trains with biologically realistic computer models of neural microcircuits. This approach also suggests a radically new approach towards neuromorphic engineering: Look directly for efficient hardware implementations of adaptive liquid state machines in order to build devices for real-time processing of sensory inputs that capture aspects of the organisation of neural computation. Learning issues in the context of this model (especially biologically plausible algorithms for unsupervised learning and applications of reinforcement learning) are topics of current research. ------------------------------------------------------------------------------------ This paper is online available (PDF, 243 KB) as # 130 from http://www.igi.tugraz.at/maass/publications.html From xwu at gauss.Mines.EDU Sun Sep 16 11:49:41 2001 From: xwu at gauss.Mines.EDU (Xindong Wu) Date: Sun, 16 Sep 2001 09:49:41 -0600 (MDT) Subject: Knowledge and Information Systems: 3(3) and 3(4), 2001 Message-ID: <200109161549.JAA08567@gauss.Mines.EDU> Knowledge and Information Systems: An International Journal ----------------------------------------------------------- ISSN: 0219-1377 (printed version) ISSN: 0219-3116 (electronic version) by Springer-Verlag Home Page: http://www.cs.uvm.edu/~xwu/kais.html =============================================== I. Volume 3, Number 3 (August 2001) ----------------------------------- http://link.springer-ny.com/link/service/journals/10115/tocs/t1003003.htm Regular Papers - Eamonn Keogh, Kaushik Chakrabarti, Michael Pazzani, Sharad Mehrotra: Dimensionality Reduction for Fast Similarity Search in Large Time Series Databases, 263-286 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030263.htm - Juan C. Augusto, Guillermo R. Simari: Temporal Defeasible Reasoning, 287-318 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030287.htm - Mei-Ling Shyu, Shu-Ching Chen, R. L. Kashyap: Generalized Affinity-Based Association Rule Mining for Multimedia Database Queries, 319-337 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030319.htm - Guanling Lee, K. L. Lee, Arbee L. P. Chen: Efficient Graph-Based Algorithms for Discovering and Maintaining Association Rules in Large Databases, 338-355 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030338.htm - Yong S. Choi: Discovering Text Databases with Neural Nets, 356-373 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030356.htm Short Papers - Ujjwal Maulik, Sanghamitra Bandyopadhyay, John C. Trinder: SAFE: An Efficient Feature Extraction Technique, 374-387 URL: http://link.springer.de/link/service/journals/10115/bibs/1003003/10030374.htm Announcements - Peter Debye Prize 2002 on Knowledge Engineering II. Volume 3, Number 4 (November 2001) -------------------------------------- http://www.cs.uvm.edu/~xwu/kais/Vol-3-4.shtml Selected and Revised Papers from KDD-2000 Workshop on Distributed and Parallel Knowledge Discovery - Distributed Web Log Mining Using Maximal Large Itemsets by Mehmet Sayal and Peter Scheuermann - Parallel and Sequential Algorithms for Data Mining Using Inductive Logic by David B. Skillicorn and Yu Wang - Distributed Clustering Using Collective Principal Component Analysis by Hillol Kargupta, Weiyun Huang, Krishnamoorthy Sivakumar, and Erik Johnson - Cost Complexity-Based Pruning of Ensemble Classifiers by Andreas L. Prodromidis and Salvatore J. Stolfo Regular Papers - Arbitrating Among Competing Classifiers Using Learned Referees by Julio Ortega, Moshe Koppel, and Shlomo Argamon-Engelson - Multivariate Discretization for Set Mining by Stephen D. Bay Call for Papers - ICDM '02: The 2002 IEEE International Conference on Data Mining (pending approval), Maebashi TERRSA, Maebashi City, Japan, November 26 - 29, 2002 2001 KAIS Reviewers Author Index From becker at meitner.psychology.mcmaster.ca Mon Sep 17 10:54:47 2001 From: becker at meitner.psychology.mcmaster.ca (Sue Becker) Date: Mon, 17 Sep 2001 10:54:47 -0400 (EDT) Subject: postdoctoral positions in neural computation Message-ID: Dear all, I would like to announce two postdoctoral training opportunities, described below. The first has more of a computational neuroscience focus: to explore neuromodulator effects in learning and memory. The second has more of a neuro-signal-processing focus: to develop artificial hearing aid technology. Can you kindly bring these to the attention of any suitable candidates. I apologize if you receive multiple copies of this posting. cheers, Sue -- Sue Becker, Associate Professor Department of Psychology, McMaster University becker at mcmaster.ca 1280 Main Street West, Hamilton, Ont. L8S 4K1 Fax: (905)529-6225 www.science.mcmaster.ca/Psychology/sb.html Tel: 525-9140 ext. 23020 --------------------------------------------------------------------------- COMPUTATIONAL AND BEHAVIOURAL NEUROSCIENCE POSTDOCTORAL POSITION A unique, multi-disciplinary postdoctoral training opportunity is available to investigate the action of neurotransmitters using computational and animal models. Topics of interest include the neuromodulatory actions of dopamine in motivated behaviours, development of fears and paranoias in hyperdopaminergic conditions, learning aversive and emotional conditioned responses, and the biological bases of emotional memory formation in structures including the hippocampus and amygdala. The candidate must have a PhD in cognitive science, computer science, or a related discipline, expertise in neural network modelling, and an interest in running and overseeing learning and memory experiments involving behavioural pharmacology. Experience with animal experimentation would be an asset but is not essential, as training will be provided. Depending upon the interests of the candidate, opportunities also exist to acquire training in human functional neuroimaging, and conduct studies with clinical populations. The position is available for a minimum of two years. This research is part of a collaborative effort involving Dr. S. Becker, Department of Psychology, McMaster University (computational neuroscience), Dr. S. Kapur, Centre for Addiction and Mental Health (CAMH) and Department of Psychiatry, University of Toronto (behavioural pharmacology, human neuroimaging with PET and fMRI) and Dr. P. Fletcher, Department of Psychology, University of Toronto, and CAMH (animal models and behavioural pharmacological studies). This position will be based at both Mcmaster University in Hamilton and the CAMH in Toronto. Interested candidates should send a letter of intention, a CV and two letters of recommendation to Dr. S. Becker at the address below. Dr. Sue Becker Department of Psychology McMaster University 1280 Main Street West, Hamilton, Ont. L8S 4K1 becker at mcmaster.ca Fax: (905)529-6225 --------------------------------------------------------------------------- POST-DOCTORAL POSITIONS IN NEURO-SIGNAL PROCESSING Funding for one or more post-doctoral fellows is available to develop and test compensation algorithms for intelligent hearing aid technology. The end product will be a wearable computing device that goes well beyond the current state of the art in hearing aid design. Topics under investigation include: beamforming algorithms, noise cancellation, modelling normal and impaired cochlear filtering, cortical feedback in auditory processing, temporal processing and auditory streaming, feedback cancellation, and binaural mechanisms. A group of researchers at McMaster University received funding from NSERC, Canada for this exciting project which is being conducted in collaboration with Gennum Corporation, one of the world's largest hearing aid manufacturers. The research team, headed by Simon Haykin in Electrical and Computer Engineering (ECE), also includes Ian Bruce, currently at Johns Hopkins University and joining the McMaster ECE faculty in Jan/2002, and Sue Becker, Ron Racine, John Platt and Laurel Trainor who are faculty members in the Psychology Department, as well as several graduate students. The team's expertise spans neural modelling, signal processing, cochlear implants, neurophysiology, neural plasticity, and auditory psychophysics and neuroimaging. Preference will be given to applicants with expertise in neurobiological modelling of the auditory system or adaptive filter design or a related field. Excellent computer programming skills are essential. Please send applications, with two references, to: Prof. Simon Haykin Communications Research Laboratory McMaster Univeristy Hamilton, Ontario, Canada L8S 4K1 haykin at mcmaster.ca --------------------------------------------------------------------------- From terry at salk.edu Mon Sep 17 15:11:23 2001 From: terry at salk.edu (Terry Sejnowski) Date: Mon, 17 Sep 2001 12:11:23 -0700 (PDT) Subject: NEURAL COMPUTATION 13:10 Message-ID: <200109171911.f8HJBNv31847@purkinje.salk.edu> Neural Computation - Contents - Volume 13, Number 10 - October 1, 2001 ARTICLE Correctness of Belief Propagation in Gaussian Graphical Models of Arbitrary Topology Yair Weiss and William T. Freeman LETTERS MOSAIC Model for Sensorimotor Learning and Control Masahiko Haruno, Daniel M. Wolpert, and Mitsuo Kawato Robust Full Bayesian Learning for Radial Basis Networks Christophe Andrieu, Nando de Freitas, and Arnaud Doucet Adaptive Algorithm for Blind Separation From Noisy Time-Varying Mixtures V. Koivunen, M. Enescu, and E. Oja Evaluating Auditory Performance Limits: I. One-Parameter Discrimination Using a Computational Model for the Auditory Nerve Michael G. Heinz, H. Steven Colburn, and Laurel H. Carney Evaluating Auditory Performance Limits: II. One-Parameter Discrimination with Random-Level Variation Michael G. Heinz, H. Steven Colburn, and Laurel H. Carney Analysis and Neuronal Modeling of the Nonlinear Characteristics of a Local Cardiac Reflex in the Rat Rajanikanth Vadigepalli, Francis J. Doyle III, and James S. Schwaber Spike-Timing-Dependent Hebbian Plasticity as Temporal Difference Learning Rajesh P. N. Rao and Terrence J. Sejnowski ----- ON-LINE - http://neco.mitpress.org/ SUBSCRIPTIONS - 2001 - VOLUME 13 - 12 ISSUES USA Canada* Other Countries Student/Retired $60 $64.20 $108 Individual $88 $94.16 $136 Institution $460 $492.20 $508 * includes 7% GST MIT Press Journals, 5 Cambridge Center, Cambridge, MA 02142-9902. Tel: (617) 253-2889 FAX: (617) 577-1545 journals-orders at mit.edu ----- From mrevow at microsoft.com Mon Sep 17 12:25:28 2001 From: mrevow at microsoft.com (Michael Revow) Date: Mon, 17 Sep 2001 09:25:28 -0700 Subject: Software developer/researcher in handwriting recognition Message-ID: Microsoft Handwriting Recognition Group Software Developer/Researcher Responsibilities center on incorporating research into the Microsoft Handwriting Recognition (HR) engine. Experience in handwriting recognition technology is not required, but a solid background in Computer Science, Mathematics, or Statistics is needed, as is the desire and ability to learn new techniques from textbooks or research papers. The HR engine is being developed for a variety of languages and includes components that implement neural nets, hidden Markov models, signal processing, statistical search, n-gram statistical language modeling, and context-free grammar language modeling. Improvements to the HR engine are targeted at increasing performance and accuracy, and improving usability problems for mainstream use in applications, such as note taking into Microsoft Office or the TabletPC. Qualifications required include at least a Bachelors degree in Computer Science, Mathematics, EE/Signal Processing, or related field. An advanced degree would be desirable. A strong interest and ability to write and support your own code in C/C++ is crucial to being successful in the group. Interested candidates should submit a recent copy of their resume and indicate their availability to: Michael Revow, 1 Microsoft Way, Redmond WA, 98052, USA. or by email to: mrevow at microsoft.com. From ken at phy.ucsf.edu Tue Sep 18 01:37:47 2001 From: ken at phy.ucsf.edu (Ken Miller) Date: Mon, 17 Sep 2001 22:37:47 -0700 Subject: Paper available: Processing in Layer 4 of the Neocortical Circuit Message-ID: <15270.56747.854955.753714@coltrane.ucsf.edu> The following review paper is available from http://www.keck.ucsf.edu/~ken (click on 'publications'): Miller, K.D., D.J. Simons and D.J. Pinto (2001). ``Processing in Layer 4 of the Neocortical Circuit: New Insights From Visual and Somatosensory Cortex''. Current Opinion in Neurobiology 11, 488-497. Summary: Recent experimental and theoretical results in cat primary visual cortex (V1) and in the whisker-barrel fields of rodent primary somatosensory cortex (S1) suggest common organizing principles for layer 4, the primary recipient of sensory input from thalamus. Response tuning of layer 4 cells is largely determined by a local interplay of feedforward excitation (from thalamus) and feedforward inhibition (from layer 4 inhibitory interneurons driven by thalamus). Feedforward inhibition dominates excitation, inherits its tuning from the thalamic input, and sharpens the tuning of excitatory cells. Recurrent excitation enhances responses to effective stimuli. We review results leading to these common pictures, and also highlight remaining differences between the two systems. Ken Kenneth D. Miller telephone: (415) 476-8217 Associate Professor fax: (415) 476-4929 Dept. of Physiology, UCSF internet: ken at phy.ucsf.edu 513 Parnassus www: http://www.keck.ucsf.edu/~ken San Francisco, CA 94143-0444 From omlin at cs.sun.ac.za Tue Sep 18 09:14:34 2001 From: omlin at cs.sun.ac.za (Dr Christian W Omlin) Date: Tue, 18 Sep 2001 15:14:34 +0200 (SAST) Subject: Postdoctoral Positions Message-ID: The Department of Computer Science at the University of the West- ern Cape (South Africa) has identified the area of intelligent systems as one of its main research thrusts. We aim to position ourselves to assume a national leadership role in this emerging field. In order to further strengthen our research capacity, we invite applications for 2 POSTDOCTORAL POSITIONS for an initial 1-year period in our Intelligent Systems Group. The appointment can be extended by an additional 2 years. Areas of interest include but are not restricted to neural networks signal processing internet applications data mining intelligent networks intelligent agents computational learning theory For applicants with research interests in bioinformatics, there exist opportunities for collaboration with the South African Na- tional Bioinformatics Institute. The successful candidates may either pursue their own research programmes or may participate in existing projects. No teaching duties are expected from the postdoctoral research fellows; however, there exists the opportunity to supervise MSc/PhD students or to give a postgraduate course if he/she so desires. We offer a competitive, tax-free stipend, a friendly atmosphere, and state-of-the-art equipment. We encourage publication of re- search results and offer financial support for travel to confer- ences. The University of the Western Cape is located 20 minutes from the center of cosmopolitan Cape Town which features scenic views, sandy beaches, great opportunities for outdoor activities, na- tional parks withs stunnning fauna and flora, and plenty of arts and culture. We also have one of the most pleasant climates in the world with 300 days of sunshine and very mild winters. We are also located 20 minutes from the heart of South Africa's wineland where world-class wines are produced at numerous vinyards. Please send your CV, a list of publications, copies of two repre- sentative papers and a statement of your research interests to comlin at uwc.ac.za. Also, please have 3 referees send letters of recommendation to the same address. Deadline for applications is November 30, 2001. ----------------------------------------------------------------- Prof. Christian W. Omlin Phone: +27-21-959/3010/2967 Intelligent Systems Group Fax: +27-21-959-2577 Department of Computer Science E-mail: comlin at uwc.ac.za University of the Western Cape URL: http://www.cs.uwc.ac.za Bellville 7535 South Africa ----------------------------------------------------------------- From cindy at cns.bu.edu Tue Sep 18 14:48:20 2001 From: cindy at cns.bu.edu (Cynthia Bradford) Date: Tue, 18 Sep 2001 14:48:20 -0400 Subject: Neural Networks 14(8) Message-ID: <200109181848.OAA28253@retina.bu.edu> NEURAL NETWORKS 14(8) Contents - Volume 14, Number 8 - 2001 ------------------------------------------------------------------ NEURAL NETWORKS LETTER: Global exponential stability of delayed Hopfield neural networks Tianping Chen CONTRIBUTED ARTICLES: ***** Psychology and Cognitive Science ***** A neurodynamical model for selective visual attention using oscillators Silvia Corchs and Gustavo Deco ***** Neuroscience and Neuropsychology ***** Information processing in dendrites, I: Input pattern generalization Kevin N. Gurney Information processing in dendrites, II: Information theoretic complexity Kevin N. Gurney ***** Mathematical and Computational Analysis ***** A modified general regression neural network (MGRNN) with new, efficient training algorithms as a robust "black box" tool for data analysis Dirk Tomandl and Andreas Schober An approach to guaranteeing generalization in neural networks J. Gary Polhill and Michael K. Weir Algebraic geometrical methods for hierarchical learning machines Sumio Watanabe A hybrid learning network for shift-invariant recognition Ruye Wang A dynamical model for the analysis and acceleration of learning in feedforward networks Nikolaos Ampazis, Stavros J. Perantonis, and John G. Taylor Estimates of average complexity of neurocontrol algorithms Tomas Hrycej Fuzzylot: A novel self-organizing fuzzy-neural rule-based pilot system for automated vehicles Michel Pasquier, Chai Quek, and Mary Toh ***** Engineering and Design ***** Real time distributed processing of multiple associated pulse pattern sequences A.J.B. Travis Myopotential denoising of ECG signals using wavelet thresholding methods Vladimir Cherkassky and Steven Kilts ------------------------------------------------------------------ Electronic access: www.elsevier.com/locate/neunet/. Individuals can look up instructions, aims & scope, see news, tables of contents, etc. Those who are at institutions which subscribe to Neural Networks get access to full article text as part of the institutional subscription. Sample copies can be requested for free and back issues can be ordered through the Elsevier customer support offices: nlinfo-f at elsevier.nl usinfo-f at elsevier.com or info at elsevier.co.jp ------------------------------ INNS/ENNS/JNNS Membership includes a subscription to Neural Networks: The International (INNS), European (ENNS), and Japanese (JNNS) Neural Network Societies are associations of scientists, engineers, students, and others seeking to learn about and advance the understanding of the modeling of behavioral and brain processes, and the application of neural modeling concepts to technological problems. Membership in any of the societies includes a subscription to Neural Networks, the official journal of the societies. Application forms should be sent to all the societies you want to apply to (for example, one as a member with subscription and the other one or two as a member without subscription). The JNNS does not accept credit cards or checks; to apply to the JNNS, send in the application form and wait for instructions about remitting payment. The ENNS accepts bank orders in Swedish Crowns (SEK) or credit cards. The INNS does not invoice for payment. ---------------------------------------------------------------------------- Membership Type INNS ENNS JNNS ---------------------------------------------------------------------------- membership with $80 (regular) SEK 660 (regular) Y 13,000 (regular) Neural Networks (plus 2,000 enrollment fee) $20 (student) SEK 460 (student) Y 11,000 (student) (plus 2,000 enrollment fee) ----------------------------------------------------------------------------- membership without $30 SEK 200 not available to Neural Networks non-students (subscribe through another society) Y 5,000 (student) (plus 2,000 enrollment fee) ----------------------------------------------------------------------------- Name: _____________________________________ Title: _____________________________________ Address: _____________________________________ _____________________________________ _____________________________________ Phone: _____________________________________ Fax: _____________________________________ Email: _____________________________________ Payment: [ ] Check or money order enclosed, payable to INNS or ENNS OR [ ] Charge my VISA or MasterCard card number ____________________________ expiration date ________________________ INNS Membership 19 Mantua Road Mount Royal NJ 08061 USA 856 423 0162 (phone) 856 423 3420 (fax) innshq at talley.com http://www.inns.org ENNS Membership University of Skovde P.O. Box 408 531 28 Skovde Sweden 46 500 44 83 37 (phone) 46 500 44 83 99 (fax) enns at ida.his.se http://www.his.se/ida/enns JNNS Membership c/o Professor Takashi Nagano Faculty of Engineering Hosei University 3-7-2, Kajinocho, Koganei-shi Tokyo 184-8584 Japan 81 42 387 6350 (phone and fax) jnns at k.hosei.ac.jp http://jnns.inf.eng.tamagawa.ac.jp/home-j.html ----------------------------------------------------------------- From taketani at tensorbio.com Tue Sep 18 18:34:50 2001 From: taketani at tensorbio.com (Makoto Taketani) Date: Tue, 18 Sep 2001 15:34:50 -0700 Subject: Director of Physiology Position Available at Tensor Biosciences Message-ID: <002201c14092$21fcc0e0$0300a8c0@CORP.TENSORBIO.COM> Director of Physiology Position Tensor Biosciences Irvine, CA USA Area of expertise: Slice Physiology Tensor Biosciences is looking for a high-level slice physiologist as a Director of Physiology. The new Director of Physiology will manage Tensor's acute physiology group which currently has four members and is expected to grow further during the next year. This position will involve managing both research and contract work. In addition, the candidate will be asked to assist with various business development activities, and as the company grows, he or she will be responsible for growing their team into a department. A strong candidate for this position should have a doctorate in the biological sciences and the following characteristics: (1) an enthusiastic start-up attitude, that is, they must be very flexible and willing to work long and hard at a variety of jobs with eye towards increasing the value of their stock; (2) strong communication skills as demonstrated by the candidates writing, presentation, and teaching experience, with any business development experience being a big plus; (3) strong management skills, with an emphasis on team building; and (4) a strong background in acute brain slice physiology with pharmaceutical industry experience being a big plus. Obviously Tensor is looking for an exceptional individual to fill this opening. Tensor will pay a salary at the upper end of the industry range for such a person, and provide he or she with a very generous stock option package. In addition, the person who fills this job will have ample opportunity to write patents and papers, and develop high-level pharmaceutical industry contacts. Tensor Biosciences (www.tensorbio.com) is developing a revolutionary new kind of drug discovery technolgies based upon Panasonic's patented multi-electrode array hardware (www.med64.com) and advanced informatics software being developed in-house to differentiate drugs and predict their effects. This work is proceeding in collaboration with a laboratory of professor Gary Lynch of the University of California at Irvine. Tensor is already delivering on its first drug analysis contract and early signs suggest that its sales of testing services and the value of the company will grow very rapidly in the next year. To apply, send a letter of application and CV to Tensor at the address listed below. Applications will be considered as they are received. Contact: Heidi Merrill Human Resources Tensor Biosciences 101 Theory, Suite 250 Irvine, CA 92697 Office:(949) 258-0309 Fax:(949) 258-0321 heidimerrill at tensorbio.com ============================================= Makoto Taketani, Ph.D. Technology Development Center Matsushita Electric Corporation of America Tel: 949-258-0310; FAX: 949-258-0321 Net: taketani at med64.com http://www.med64.com ------------------------------------------ From Chris.Diehl at jhuapl.edu Tue Sep 18 09:23:31 2001 From: Chris.Diehl at jhuapl.edu (Diehl, Chris P.) Date: Tue, 18 Sep 2001 09:23:31 -0400 Subject: Ph.D. Thesis Announcement Message-ID: <91D1D51C2955D111B82B00805F1998950BD8C946@aples2.jhuapl.edu> Dear Connectionists, The following Ph.D. thesis is now available at http://www.cpdiehl.org. Toward Efficient Collaborative Classification for Distributed Video Surveillance Christopher P. Diehl Ph.D. Thesis Department of Electrical and Computer Engineering Carnegie Mellon University Abstract In this thesis, we propose a general strategy for automated video surveillance that relies on collaboration between the surveillance system and the user. Such collaboration enables the user to help the system incrementally acquire the necessary context for truly robust surveillance. The success of this strategy is dependent on the ability of the system to identify novel instances of known or unknown classes that it does not understand. This, in turn, allows the user to focus only on the observations with the highest uncertainty that require interpretation. Designing a real-time classification process that supports novelty detection is nontrivial. The real-time constraint dictates computational simplicity, whereas novelty detection requires a high dimensional feature space to aid in discriminating between the known and unknown classes. The majority of this work focuses on the problem of simultaneously satisfying these conflicting constraints. We consider these issues in the context of a relevant surveillance task and evaluate the performance of the resulting classification process in the CMU Cyberscout distributed video surveillance system. Dr. Chris Diehl System and Information Sciences Group Research and Technology Development Center Applied Physics Laboratory Johns Hopkins University 443-778-3457 (Office) 443-778-6904 (Fax) http://www.cpdiehl.org From cohn+jmlr at cs.cmu.edu Wed Sep 19 15:48:34 2001 From: cohn+jmlr at cs.cmu.edu (JMLR) Date: Wed, 19 Sep 2001 15:48:34 -0400 Subject: Two new papers in the Journal of Machine Learning Research Message-ID: The Journal of Machine Learning Research (www.jmlr.org) is pleased to announce the availability of two new papers in electronic form: Tracking the Best Linear Predictor Mark Herbster and Manfred K. Warmuth Journal of Machine Learning Research 1(Sep):281-309, 2001 Prior Knowledge and Preferential Structures in Gradient Descent Learning Algorithms Robert E. Mahony, Robert C. Williamson Journal of Machine Learning Research 1(Sep):311-355, 2001 ---------------------------------------- Tracking the Best Linear Predictor Mark Herbster and Manfred K. Warmuth Journal of Machine Learning Research 1(Sep):281-309, 2001 Abstract In most on-line learning research the total on-line loss of the algorithm is compared to the total loss of the best off-line predictor u from a comparison class of predictors. We call such bounds static bounds. The interesting feature of these bounds is that they hold for an arbitrary sequence of examples. Recently some work has been done where the predictor u_t at each trial t is allowed to change with time, and the total on-line loss of the algorithm is compared to the sum of the losses of u_t at each trial plus the total ``cost'' for shifting to successive predictors. This is to model situations in which the examples change over time, and different predictors from the comparison class are best for different segments of the sequence of examples. We call such bounds shifting bounds. They hold for arbitrary sequences of examples and arbitrary sequences of predictors. Naturally shifting bounds are much harder to prove. The only known bounds are for the case when the comparison class consists of a sequences of experts or boolean disjunctions. In this paper we develop the methodology for lifting known static bounds to the shifting case. In particular we obtain bounds when the comparison class consists of linear neurons (linear combinations of experts). Our essential technique is to project the hypothesis of the static algorithm at the end of each trial into a suitably chosen convex region. This keeps the hypothesis of the algorithm well-behaved and the static bounds can be converted to shifting bounds. ---------------------------------------- Prior Knowledge and Preferential Structures in Gradient Descent Learning Algorithms Robert E. Mahony, Robert C. Williamson Journal of Machine Learning Research 1(Sep):311-355, 2001 Abstract A family of gradient descent algorithms for learning linear functions in an online setting is considered. The family includes the classical LMS algorithm as well as new variants such as the Exponentiated Gradient (EG) algorithm due to Kivinen and Warmuth. The algorithms are based on prior distributions defined on the weight space. Techniques from differential geometry are used to develop the algorithms as gradient descent iterations with respect to the natural gradient in the Riemannian structure induced by the prior distribution. The proposed framework subsumes the notion of "link-functions". ---------------------------------------- These papers and earlier papers in Volume 1 are available electronically at http://www.jmlr.org in PostScript, PDF and HTML formats; a bound, hardcopy edition of Volume 1 will be available later this year. -David Cohn, Managing Editor, Journal of Machine Learning Research ------- This message has been sent to the mailing list "jmlr-announce at ai.mit.edu", which is maintained automatically by majordomo. To subscribe to the list, send mail to listserv at ai.mit.edu with the line "subscribe jmlr-announce" in the body; to unsubscribe send email to listserv at ai.mit.edu with the line "unsubscribe jmlr-announce" in the body. From mkm at hnc.com Wed Sep 19 23:26:12 2001 From: mkm at hnc.com (McClarin, Melissa) Date: Wed, 19 Sep 2001 20:26:12 -0700 Subject: HNC Software seeks a Senior Staff Scientist Message-ID: <72A838A51366D211B3B30008C7F4D363085AE7F7@pchnc.hnc.com> HNC Software is seeking a full time Senior Staff Scientist, a job based at our headquarters in San Diego, California. We provide Customer Insight through Intelligent Response, decision management and customer analytics software that enables companies in the financial, telecommunications, e-commerce and insurance industries to acquire, manage and retain customers. Senior Staff Scientist Job Duties/Responsibilities: General Duties: PhD with some experience in the development of predictive models to real-world business applications. Knowledge in worker compensation/auto insurance. Superior programming skills. Excellent communication skills(both writing and verbal). Specific Duties: Design and implement innovative solutions to a wide variety of technical problems in insurance industry using data mining techniques, statistics and neural network modeling. As a technical lead of projects and product development, contribute to all phases of model development which including data analysis, solution design, model training, and presentation of results. Provides technical guidance to other team members. Works with other functional groups to ensure integrity of deliverable. Required Qualifications: MS/PhD in computer science, statistics, math, Engineering, or related field. Demonstrated experience applying advanced statistics, data mining techniques to real world data. Must be a strong problem solver and fast learner, possessing excellent analytical abilities. Ability to work independently and provides team technical guidance. Comfortable with a variety statistical packages and computer systems, such as SAS, C, Unix, Windows. Preferred Qualifications: PhD with 5 years experience in the development of predictive models to real-world business applications. Knowledge in worker compensation/auto insurance, and software development. Superior programming skills. Excellent communication skills(both writing and verbal). Experience as a team leader To apply, please use one of the following methods: Mail: 5935 Cornerstone Ct. W San Diego, CA 92121 Email: hireme at hnc.com Online: http://www.hnc.com/careers_12/index - job number 366 From schrott at in.tum.de Thu Sep 20 03:50:46 2001 From: schrott at in.tum.de (Dr. Gerhard Schrott) Date: Thu, 20 Sep 2001 09:50:46 +0200 Subject: Research or Post-Doc Research Position at TU Munich Message-ID: <3BA99FD6.F99F24C0@in.tum.de> Technical University of Munich Department of Computer Science The Department of Computer Science, Robotics and Embedded Systems unit, invites applications for a Research or Post-Doc Research Position within the EU-funded long-term research programme "Neuroinformatics for Living Artefacts" in a pan-European research project aiming at exploring the fundamentals and the potential of Imitation Learning for future robot systems. Applications would be particularly welcome from candidates with research interests (and a record of innovative research results) in one or more of the following areas: robotics (e.g. Sensor data processing and sensor data fusion; Robust vision/tracking for robots; Learning and adaptive control; Behaviour engineering and architectures) and/or neuro-modelling (e.g. modelling of dynamical systems; Optimization and nonlinear dynamics; Coupling of motor behaviour and visual perception; pattern generation for motion and postural control; Transfer of the mathematical concepts into algorithms). The successful candidate will have a graduate or a postgraduate qualification - ideally to doctoral level - in computer science or neurosciences, together with practical experience in robotics. Further details may be obtained from Prof. A. Knoll, knoll at in.tum.de. Closing date: 5 October 2001. Applications including the CV, certificates, publication list should be sent to: Technische Universitaet Muenchen, Fakultaet fuer Informatik, LS VI - Robotics and Embedded Systems, z.H. R. te Vehne, Orleansstrasse 34, 80667 Muenchen, Germany, or by E-mail to Vehne at in.tum.de From bengio at idiap.ch Fri Sep 21 07:57:49 2001 From: bengio at idiap.ch (Samy Bengio) Date: Fri, 21 Sep 2001 13:57:49 +0200 (MEST) Subject: Torch: a new machine learning library in C++/GPL Message-ID: Dear connectionists people, We would like to announce the availability of yet another machine learning library, written in C++ under the GPL license. This new library is named "Torch" and has been written mainly by Ronan Collobert, also known for his widely used SVMTorch package for SVMs. Other contributors include Samy Bengio and Johnny Mari=E9thoz. Currently, the main features of Torch are the following: - A lot of things in gradient-machines, that is, machines which could be learned with gradient descent. This includes Multi-Layered Perceptrons, Radial Basis Functions and Mixtures of Experts. In fact there are a lot of small "modules" available (Linear, Tanh, SoftMax...) that you can plug as you want to get what you want. - Support Vector Machine, in classification and regression (mostly the same code as in SVMTorch but now integrated in the Torch library). - A Distribution package which includes for the moment Kmeans, Gaussian Mixture Models (GMMs) and Hidden Markov Models (HMMs). - A few non-parametric models such as K-nearest-neighbors, Parzen Regression and Parzen Density Estimator. - Tools to do either "train/test" experiments or K-fold cross-validation - A lot of measurers (to print for instance the mean squared error or the classification error of as many datasets as wanted, during training or testing). - A few typical "main.cc" examples to understand how to create your own experiments such as mlp.cc (train and test an MLP), gmm.cc (maximize the likelihood of a GMM), hmm.cc (maximize the likelihood of an HMM), mixture_softmax (train and test a mixture of experts). This library is intented to provide the state-of-the-art of the best algorithms in machine learning. Therefore, if you know C++, are working in machine learning, and want to develop your own algorithms or use well-known machine learning algorithms, Torch is for you! Visit the official Torch website at http://www.torch.ch to know more about Torch and/or download it. Note that, of course, Torch is and will always be under development... Note also that it was designed for Unix and Linux systems... Have fun! ----- Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From crocker at CoLi.Uni-SB.DE Tue Sep 25 04:31:05 2001 From: crocker at CoLi.Uni-SB.DE (Matthew Crocker) Date: Tue, 25 Sep 2001 10:31:05 +0200 Subject: Positions in Language Processing In-Reply-To: <200108311318.PAA07149@top.coli.uni-sb.de> Message-ID: <200109250834.KAA07734@top.coli.uni-sb.de> Two Research Positions Available in Computational and Experimental Psycholinguistics, Saarbruecken, Germany The Department of Computational Linguistics, Saarland University is seeking to fill two research positions in the areas of Computational and Experimental Psycholinguistics. Persons taking up the positions will be involved in a newly funded project entitled: "Adaptive Mechanisms for Human Language Processing". The project aims to develop wide-coverage, probabilistic models of human language processing, as informed by evidence obtained from large corpora, and both on-line and off-line psycholinguistic experiments. 1. Computational Linguist: experience in the development of probabilistic models of language processing. In particular, we are interested in developing incremental models of syntactic and semantic processing. Knowledge of one or more of the following is desirable: probabilistic parsing techniques, machine learning of natural language, connectionist language modelling, and experience working with large corpora. 2. Experimental Psycholinguist: experience in designing, running and analysing psycholinguistic experiments, and familiarity with (some of) the following paradigms: self-paced reading, eye-tracking (fixed or head-mounted), language production, and web-based experiments. Candidates should have research experience in a relevant subject area, and ideally will hold a PhD. The position is on the BAT IIa scale (roughly up to DM 80K per annum, depending on age and family status) and is tenable for 3 years with the possibility of renewal. The positions are available from January 1st, 2002. Applications received before November 1, 2001 are assured fullest consideration. Interested persons should send a letter of application giving contact details for three possible referees and a full CV to (e-mail applications are also welcome): Prof. Dr. Matthew W. Crocker Psycholinguistics Group, Gebaeude 17 Department of Computational Linguistics Saarland University 66041 Saarbruecken, Germany E-mail: crocker at coli.uni-sb.de The Department offers state of the art research facilities including head-mounted and DPI eye-tracking equipment, powerful Unix servers for statistical modeling, and an extensive corpus infrastructure. Saarland University has an international profile in computational linguistics, cognitive science and computer science. Research is supported by a European Graduate School for "Language Technology and Cognitive Systems" (joint with Edinburgh University) and a Centre of Excellence (SFB 378) in the area of "Resource Adaptive Cognitive Processes". The University of Saarland seeks to increase the proportion of women in positions where they are under-represented, and therefore particularly encourages applications from women. In the selection procedure, disabled persons with equivalent qualifications will be favoured. From: Martin Giese Subject: PhD position in computer vision / biomedical engineering From bogus@does.not.exist.com Wed Sep 26 17:39:05 2001 From: bogus@does.not.exist.com () Date: Wed, 26 Sep 2001 23:39:05 +0200 Subject: No subject Message-ID: The Laboratory for Action Representation and Learning at the Department of Cognitive Neurology and the Max-Planck Institute for Biological Cybernetics in Tuebingen (Germany) offers a PhD position for students in engineering, computer science, physics or mathematics. Aim of the project is the development of technical methods for the analysis of complex movements of neurological patients. On the theory side, the project focuses on the development of new learning techniques for the representation of complex movements. The practical side of the project includes data acquisition with modern motion capture systems, and the analysis and visualization of the data using methods from computer vision and computer graphics. In addition, the successful candidate should be willing to communicate with neurologists and patients. The project will be realized in close collaboration with the Department for Neurology at the Universitaetsklinik Tuebingen. An interesting environment for computer vision and machine learning is provided by the divisions for Virtual Reality and machine learning reasearch at the Max-Planck-Institute. In addition, the group has close contacts with the Center for Biological and Computational Learning at M.I.T. The ideal candidate has a good computational background, and some experience in computer vision / graphics, machine learning, or robotics. Payment will be, dependent on prior experience, BAT IIa / 2 or BAT IIa for 3 years (extendable). Starting date should be as soon as posible. The position is funded by the German Volkswagen Foundation. For further information please contact: Dr. Martin Giese Laboratory for Action Representation and Learning Max-Planck-Institut for Biological Cybernetics Spemannstr. 34 D-72076 Tuebingen GERMANY Email: martin.giese at tuebingen.mpg.de Applicants are asked to submit their CV, a bibliography, and the names of two references. Applications and references should be sent by email to the same address. -------------------------------------------------------- Dr. Martin Giese Laboratory for Action Representation and Learning Max-Planck-Institut for Biological Cybernetics Spemannstr. 34 D-72076 Tuebingen GERMANY Tel.: +49 7071 601 724 Fax: +49 7071 601 616 Email: martin.giese at tuebingen.mpg.de -------------------------------------------------------- From murphyk at cs.berkeley.edu Wed Sep 26 22:29:18 2001 From: murphyk at cs.berkeley.edu (Kevin Murphy) Date: Wed, 26 Sep 2001 19:29:18 -0700 Subject: paper on BNT now available Message-ID: <3BB28EFE.236528D0@cs.berkeley.edu> I am pleased to announce the following paper "The Bayes Net Toolbox for Matlab", by Kevin Murphy. Published in: Computing Science and Statistics: Proceedings of Interface, volume 33, 2001 Available at http://HTTP.CS.Berkeley.EDU/~murphyk/Papers/bnt.ps.gz http://HTTP.CS.Berkeley.EDU/~murphyk/Papers/bnt.pdf Abstract: The Bayes Net Toolbox (BNT) is an open-source Matlab package for directed graphical models. BNT supports many kinds of nodes (probability distributions), exact and approximate inference, parameter and structure learning, and static and dynamic models. BNT is widely used in teaching and research: the web page has received over 28,000 hits since May 2000. In this paper, we discuss a broad spectrum of issues related to graphical models (directed and undirected), and describe, at a high-level, how BNT was designed to cope with them. We also compare BNT to other software packages for graphical models, and to the nascent OpenBayes effort. From strom at ece.ogi.edu Wed Sep 26 20:37:23 2001 From: strom at ece.ogi.edu (Dan Hammerstrom) Date: Wed, 26 Sep 2001 17:37:23 -0700 Subject: Post-Doc position Message-ID: <4.2.0.58.20010926173522.01d68ef8@spruce.ece.ogi.edu> Postdoctoral Position in Computational Modelling for Sensorimotor Control A postdoctoral training opportunity is available to investigate computational models for sensorimotor control with application to autonomous robot navigation. The research aims at developing models for bottom-up/top-down sensory integration and high-level representations for visuomotor control. The current investigations address sparse representations and on-line learning within such representations, working memory and goal memory representations and learning; representation of risk, risk anticipation and risk aversion. The candidate must have a PhD in cognitive science, computer science/engineering or a related discipline, expertise in one or more of the following: neural modelling, learning algorithms, machine vision, roving robot programming. This research is part of a collaborative effort with funding from NASA under its Revolutionary Computing Program and a Gordon & Betty Moore Research Scholarship. The research involves Prof. Marwan Jabri (OHSU), Prof. Dan Hammerstrom (OHSU) and Prof. Terrence Sejnowski (Salk). The present position is located at the OGI School of Science and Engineering, OHSU West Campus, Beaverton, Oregon. Interested candidates should send a letter of intention, a CV and two letters of recommendation to Prof. M. Jabri at the address below. Marwan Jabri OGI School of Science & Engineering Oregon Health and Sciences University 20000 N.W. Walker Rd. Beaverton, OR 97006, USA Tel: (+1-503) 748-7274 Fax: (+1-503) 748-4070 marwan at ece.ogi.edu From duch at phys.uni.torun.pl Wed Sep 26 10:41:10 2001 From: duch at phys.uni.torun.pl (Wlodzislaw Duch) Date: Wed, 26 Sep 2001 16:41:10 +0200 Subject: ICNNSC 2002, Sixth International Conference on Neural Networks and Soft Computing, call for papers Message-ID: ICNNSC 2002, http://www.icnnsc.pcz.czest.pl Sixth International Conference on Neural Networks and Soft Computing June 11-15, 2002, Zakopane, Poland organised by Polish Neural Networks Society in cooperation with IEEE Neural Networks Council. Honorary chairs Lotfi Zadeh - USA Jacek Zurada - USA General co-chairs Wlodzislaw Duch - Poland Janusz Kacprzyk - Poland Leszek Rutkowski - Poland Ryszard Tadeusiewicz - Poland Zdzislaw Pawlak - Poland The Sixth International Conference on Neural Networks and Soft Computing ICNNSC 2002 will be held in Zakopane (situated in the HighTatra mountains), Poland on 11-15 June, 2002. The conference will provide an excellent opportunity for scientist and engineers to present and discuss the latest scientific results and methods. The conference will include keynote addresses, contributed papers, and numerous lectures and tutorials on a wide range of topics. Important Dates: 1. Submission of papers in accordance with Springer guidelines (6 pages max.): February 15, 2002 2. Notification of acceptance: March 30, 2002 3. Submission of camera-ready papers May5, 2002 4. Conference date: 11-15 June, 2002 Scope The conference covers all topics in neural networks, fuzzy systems, evolutionary computation, hybrid methods, including but not limited to: Supervised and unsupervised learning Neural network theory Neural network architectures Hardware implementations Fuzzy logic Fuzzy optimisation Fuzzy control Fuzzy computing with words Theory of evolutionary algorithms Evolutionary design Evolutionary scheduling and optimisation Rough sets theory: foundations and applications Multi-agent systems Data mining Applications: prediction, identification, pattern recognition, image and signal processing, speech and computer vision, financial engineering and forecasting, medicine, industry ... The working language of the conference is English. Only original, unpublished papers in the aforementioned fields are invited. Authors should submit 3 hard copies of full papers (up to six pages) to the conference office and an electronic version to: icnnsc at kik.pcz.czest.pl before February 15, 2002. Paper Submission and Publication The papers should be organized in accordance with a common scientific structure (abstract, state of the art in the field, intention, used methodology, obtained results and references). Papers will be refereed by an international committee, and accepted on the basis of their scientific merit and relevance to the conference topics. After the notification of acceptance (March 30, 2002), authors will be allowed to make a correction in accordance with the suggestions of the reviewers and submit final camera-ready papers before May 5, 2002. Abstracts of accepted papers will be available during the conference in the form of a brochure. The conference proceedings will be published in the Springer-Verlag series "Advances in Soft Computing" and distributed among the participants after the conference. Accepted papers must be presented by author(s) personally to be published in the conference proceedings. Venue The conference is organized in Poland's premier mountain resort Zakopane, the beautiful and picturesque capital of the Polish High Tatra Mountains. Even though Zakopane is called Poland's winter capital, it is an excellent place to visit in spring. Situated at the foot of the Tatra Mountains and the Tatra National Park (with Poland's highest peak, Rysy mountain (2499m)), it offers you a possibility to hike on over 240km of marked hiking trails for beginners as well as experienced climbers and explore caves accessible only in summer. The town is filled with historic monuments, museums, art galleries and exhibitions. On your way to Zakopane you may visit Cracow, one of the most historic Polish cities, with a wonderfully preserved old city center and the largest medieval square in Europe. For details please visit the ICNNSC2002 Web page http://www.icnnsc.pcz.czest.pl From l.s.smith at cs.stir.ac.uk Thu Sep 27 04:29:15 2001 From: l.s.smith at cs.stir.ac.uk (Leslie Smith) Date: Thu, 27 Sep 2001 09:29:15 +0100 Subject: Research posts in Scotland: INCITE Message-ID: <3BB2E35B.F6CA59CB@cs.stir.ac.uk> New Institute based at the University of Stirling, Scotland, with research fellow positions in Stirling, Edinburgh and Paisley Universities, Scotland: Institute for Neuronal Computational Intelligence and Technology (INCITE) INCITE aims to design systems which combine the speed of silicon with the robustness of human performance. This will underpin the next generation of IT systems, enabling them to approach the complexity of human performance. INCITE is a newly funded Institute integrating expertise from the Universities of Stirling, Edinburgh and Paisley on neuromorphic engineering, robotics, neural networks and cognitive and computational neuroscience. INCITE will combine research and technology transfer functions. The INCITE web page is at http://www.cn.stir.ac.uk/incite/, and contact details can be found from there. Research Fellow (3 posts) INCITE will perform research at Stirling, Paisley and Edinburgh Universities. One Research Fellow will be appointed at each location, to assist with ongoing research. For more information about each location, see the www page above. The posts are for three years, starting after 1 Feb, 2002. Salary will be within the UK Universities Research & Analogous Staff Scale Grade 1A (UK?17,451-26,229 p.a.). Quote Reference 5368S/413 (Stirling), 5368E/413 (Edinburgh) and 5368P/413 (Paisley) For informal discussion prospective applicants may phone Professor W?rg?tter (+(44) 1786 466369) or Professor Smith (+(44) 1786 467435). Further particulars for all posts are available from the Personnel Office, University of Stirling, Stirling, FK9 4LA, tel: (+(44) 1786 467028, fax (+(44) 1786) 466155 or email personnel at stir.ac.uk quoting the appropriate reference number. Closing date for applications: Wednesday, 10 October 2001. www.personnel.stir.ac.uk AN EQUAL OPPORTUNITIES EMPLOYER -- Professor Leslie S. Smith, Head of Department, Dept of Computing Science and Mathematics, University of Stirling, Stirling FK9 4LA, Scotland l.s.smith at cs.stir.ac.uk Tel (44) 1786 467435 Fax (44) 1786 464551 www http://www.cs.stir.ac.uk/~lss/ From shahar at discus.anu.edu.au Thu Sep 27 23:50:48 2001 From: shahar at discus.anu.edu.au (Shahar Mendelson) Date: Fri, 28 Sep 2001 05:50:48 +0200 Subject: Machine Learning Summer School Message-ID: <4.2.2.20010928055015.00b63d00@in.zahav.net.il> Machine Learning Summer School We would like to inform you that The Australian National University will be hosting a Machine Learning Summer School. The school will consist of six short courses and a series of special invited talks, taught by experts from Australia and overseas. The school will be held between February 11 and February 22, 2002. It is suitable for all levels, both for people without previous knowledge in Machine Learning and those wishing to broaden their expertise in this area. It will allow the participants to get in touch with international experts in this field. The list of topics and speakers includes: - Reinforcement Learning Peter Bartlett, Biowulf Technologies - Boosting Ron Meir, Technion - Statistical Learning Theory Shahar Mendelson, ANU - Online Learning and Bregmann divergences Gunnar Ratsch, ANU - Support Vector Machines Bernhard Schoelkopf, Biowulf Technologies and MPIK - Bayesian Kernel Methods Alex Smola, ANU We are offering a limited number of scholarships for Australian students with a strong academic background. Students who are interested should include their CV with their application. The registration cost of the school is $1,200 per person for participants from industry and $240 per person for academics. Students are eligible for a further discount and may register for $120 per person. All prices are in Australian dollars and include GST. The closing date for early registration is December 31, 2001. Registration after this date will be subject to 33% surcharge. The closing date for scholarship applications is December 7, 2001. For further information, visit our website at http://mlg.anu.edu.au/summer2002 or send e-mail to Michelle.Moravec at anu.edu.au or Diane.Kossatz.anu.edu.au. Regards, Alex Smola and Shahar Mendelson From tleen at cse.ogi.edu Fri Sep 28 19:30:54 2001 From: tleen at cse.ogi.edu (Todd Leen) Date: Fri, 28 Sep 2001 16:30:54 -0700 Subject: Postdoctoral Position in Neural Modeling Message-ID: <3BB5082D.CE1652F1@cse.ogi.edu> Postdoctoral Research Position in Theoretical and Computational Neuroscience Oregon Health & Science University, Portland, Oregon. A unique postdoctoral position is available at the Oregon Health & Science University. Dr. Todd Leen of the Oregon Graduate Institute (OGI), and Dr. Patrick Roberts of the Neurological Sciences Institute (NSI) invite applications for a senior research associate. This NSF-funded project will investigate the sources and effects of noise in the electrosensory systems of mormyrid fish. The mormyrid is a weakly electric fish that uses is electrosensory systems for navigation, communication, and hunting. The system is experimentally well-characterized physiologically and anatomically from synaptic plasticity up through the behaving animal; much of this experimental work was and is carried out by Dr. Curtis Bell at NSI. Previous modeling work on this system is featured on Dr. Robert's page http://www.ohsu.edu/nsi/faculty/robertpa/robertpa.htm. This project will focus on stochastic dynamics of adaptation and signal processing in the mormyrid. The successful candidate will have background in statistical or theoretical physics or in neural systems modeling with facility in theory as well as computation, and a PhD in either theoretical physics, theoretical neurobiology, or an allied field. The position is open immediately. Candidates should send a CV and arrange to have three letters of recommendation sent to Todd K. Leen, Oregon Graduate Institute, 20000 NW Walker Road, Beaverton, OR 97006, or by email to tleen at cse.ogi.edu. Oregon Health & Science University is an Affirmative Action / Equal Opportunity Employer.