From achim at cse.unsw.edu.au Wed Dec 1 03:57:50 1999 From: achim at cse.unsw.edu.au (Achim Hoffmann) Date: Wed, 1 Dec 99 19:57:50 +1100 Subject: Book Announcement: Paradigms of Artificial Intelligence, Springer-Verlag Message-ID: <991201085750.15517@cse.unsw.edu.au> My apologies, if you receive this announcement more than once. Paradigms of Artificial Intelligence A Methodological and Computational Analysis by Achim Hoffmann, Springer-Verlag 1998 http://www.cse.unsw.edu.au/~achim/Book/ From aida at kouku-k.ac.jp Thu Dec 2 03:20:13 1999 From: aida at kouku-k.ac.jp (Toshiaki Aida) Date: Thu, 02 Dec 1999 17:20:13 +0900 Subject: Paper Available Message-ID: <38462BBD.C6E43C05@kouku-k.ac.jp> Dear colleagues, The following paper is now available at http://xxx.lanl.gov/abs/cond-mat/9911474 , in which we discussed the control of bin size for optimal on-line learning of probability distributions. TITLE: Field Theoretical Analysis of On-line Learning of Probability Distributions, Toshiaki Aida, Physical Review Letters 83 (1999) 3554-3557. ABSTRACT: On-line learning of probability distributions is analyzed from the field theoretical point of view. We can obtain an optimal on-line learning algorithm, since renormalization group enables us to control the number of degrees of freedom of a system according to the number of examples. We do not learn parameters of a model, but probability distributions themselves. Therefore, the algorithm requires no a priori knowledge of a model. Best regards, Toshiaki Aida Department of Physics, Tokyo Institute of Technology, Department of Aeronautics, Tokyo Metropolitan College of Aeronautical Engineering, E-mail: aida at stat.phys.titech.ac.jp From maarten at swi.psy.uva.nl Thu Dec 2 07:39:19 1999 From: maarten at swi.psy.uva.nl (Maarten van Someren) Date: Thu, 2 Dec 1999 13:39:19 +0100 (MET) Subject: MLNET - for your mailinglist Message-ID: MLNET - a European network of excellence in Machine Learning, Case-Based Reasoning and Knowledge Acquisition MLNET is a "network of excellence" in Machine Learning that is funded by the European Community to coordinate research, development and application of Machine Learning, Case-Based Reasoning and Knowledge Acquisition in the European Community. Activities of the network are: * development and maintenance of an electronic information service that gives access to information on: research groups, projects, datasets, software, educational materials and links to other information services (see http://www.mlnet.org/). This is the successor to the GMD information service developed earlier by MLNET. * development and maintenance of a scientific and technological outlook ("technological roadmap") for these areas that acts as a guideline for the planning and coordination of research and development. (Members of MLNET can take part in developing this.) * development and maintenance of educational and "advertising" materials for the areas of the network. * organisation and support of events (e.g. European Conference on Machine Learning, European Workshop on Case-Based Reasoning, European Knowledge Acquisition Workshop, Multi-strategy learning workshop). Members of MLNET can propose to organise such events and apply for support. * funding of "ambassadors" who give presentations on Machine Learning, Case-Based Reasoning or Knowledge Acquisition OUTSIDE the areas. (proposals for "ambassador visits" by a member of MLNET can be submitted to the coordinator of MLNET, see below). Institutes and companies in the European community and associated states can apply for membership. At the moment the information service is available for everyone. Only members of MLNET can receive financial support for organising or visiting MLNET events. MLNET collaborates with three other European networks (ERUDIT on uncertainty modelling, EvoNet on evolutionary computing and NEuroNet on neural network computing) on the theme Computational Intelligence and Learning (see http://www.dcs.napier/coil/). More information can be found at the MLNET Information Service. - maarten van someren (coordinator of MLNET, maarten at swi.psy.uva.nl) From rsun at cecs.missouri.edu Sat Dec 4 19:12:56 1999 From: rsun at cecs.missouri.edu (Ron Sun) Date: Sat, 4 Dec 1999 18:12:56 -0600 Subject: Papers on cognitive modeling using reinforcement learning Message-ID: <199912050012.SAA05951@pc113.cecs.missouri.edu> Announcing four papers on cognitive modeling and cognitive architectures based on hybrid reinforcement learning methods --- the CLARION model: A paper on cognitive modeling using CLARION: -------------------------------------------------- From piuri at elet.polimi.it Sat Dec 4 11:04:14 1999 From: piuri at elet.polimi.it (Vincenzo Piuri) Date: Sat, 04 Dec 1999 17:04:14 +0100 Subject: IJCNN'2000: CFP & Submission of extended versions to IEEE Transactions on Neural Networks!!!! Message-ID: <3.0.5.32.19991204170414.0151ac60@elet.polimi.it> ======================================================================== IEEE-INNS-ENNS INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS Grand Hotel di Como, Como, Italy - 24-27 July 2000 PAPER SUBMISSION DEADLINE is coming soon!!!! Please, have a look to the call for papers in the conference web site. Do not miss the opportunity to submit a paper and participate actively to the conference! CALL FOR SPECIAL SESSIONS AND TUTORIALS have been also published. You can find them on the conference web site. NEW!!!! NEW!!!! NEW!!!!! SUBMISSION OF AN EXTENDED VERSION OF THE PAPER TO IEEE-TNN: The authors of selected IJCNN'2000 papers will be invited to submit an extended version to the IEEE Transactions on Neural Networks by 30 SEPTEMBER 2000. These submitted papers will undergo the usual Transactions review process. However, the Transactions Editors will receive the IJCNN'2000 reviewing reports for each selected IJCNN'2000 paper, and thus the reviewing process is likely to be simpler and faster, although accurate as usual. For each paper, the Associate Editor in charge may in fact ask the IJCNN'2000 reviewers (and/or PC members) to provide an in-depth review of the extended version for the IEEE-TNN. Since the reviewers and the PC member had already evaluated a preliminary version of the paper, their work is likely to be simplified. official conference web site: http://www.ims.unico.it/2000ijcnn.html conference mirror web site: http://www.lans.ece.utexas.edu/2000ijcnn.html ====================================================================== Vincenzo Piuri Department of Electronics and Information, Politecnico di Milano piazza L. da Vinci 32, 20133 Milano, Italy phone +39-02-2399-3606 secretary +39-02-2399-3623 fax +39-02-2399-3411 email piuri at elet.polimi.it From nnsp00 at neuro.kuleuven.ac.be Fri Dec 3 11:27:07 1999 From: nnsp00 at neuro.kuleuven.ac.be (NNSP2000, Sydney) Date: Fri, 03 Dec 1999 17:27:07 +0100 Subject: IEEE workshop on Neural Networks for Signal Processing (NNSP), Sydney, Australia, December 2000. Message-ID: <3847EF5B.CF3908CB@neuro.kuleuven.ac.be> Please find attached advanced information about the IEEE workshop on Neural Networks for Signal Processing (NNSP) which is to be held in Sydney, Australia, in December 2000. In case you would like to be removed from our mailing list: reply to this mail with as subject "remove". Marc M. Van Hulle Katholieke Universiteit Leuven Belgium *********************************************** **** CALL FOR PAPERS **** **** submission deadline: March 31, 2000 **** *********************************************** December 11-13, 2000, Sydney, Australia NNSP'2000 homepage: http://eivind.imm.dtu.dk/nnsp2000 Thanks to the sponsorship of IEEE Signal Processing Society and the IEEE Neural Networks Council the tenth of a series of IEEE workshops on Neural Networks for Signal Processing will be held at the University of Sydney Campus, Sydney, Australia. The workshop will feature keynote lectures, technical presentations, and panel discussions. Papers are solicited for, but not limited to, the following areas: Algorithm and Architectures: Artificial neural networks (ANN), adaptive signal processing, Bayesian modeling, MCMC, generalization, design algorithms, optimization, parameter estimation, nonlinear signal processing, Markov models, fuzzy systems (FS), evolutionary computation (EC), synergistic models of ANN/FS/EC, and wavelets. Applications: Speech processing, image processing, sonar and radar, data fusion, intelligent multimedia and web processing, OCR, robotics, adaptive filtering, blind source separation, communications, sensors, system identification, and other general signal processing and pattern recognition applications. Implementations: Parallel and distributed implementation, hardware design, and other general implementation technologies. PAPER SUBMISSION PROCEDURE Prospective authors are invited to submit a full paper using the electronic submission procedure described at the workshop homepage: http://eivind.imm.dtu.dk/nnsp2000 Accepted papers will be published in a hard-bound volume by IEEE and distributed at the workshop. SCHEDULE Submission of full paper: March 31, 2000 Notification of acceptance: May 31, 2000 Submission of photo-ready accepted paper: July 15, 2000 Advanced registration, before: September 15, 2000 ORGANIZATION Honorary Chair Bernard WIDROW Stanford University General Chairs Ling GUAN University of Sydney email: ling at ee.usyd.edu.au Kuldip PALIWA Griffith University email: kkp at shiva2.me.gu.edu.au Program Chairs Tulay Adali University of Maryland, Baltimore County email: adali at umbc.edu Jan LARSEN Technical University of Denmark email: jl at imm.dtu.dk Finance Chair Raymond Hau-San WONG University of Sydney email: hswong at ee.usyd.edu.au Proceedings Chairs Elizabeth J. WILSON Raytheon Co. email: bwilson at ed.ray.com Scott C. DOUGLAS Southern Methodist University email: douglas at seas.smu.edu Publicity Chair Marc Van HULLE Katholieke Universiteit, Leuven email: marc at neuro.kuleuven.ac.be Registration and Local Arrangements Stuart PERRY Defence Science and Technology Organisation email: Stuart.Perry at dsto.defence.gov.au Europe Liaison Jean-Francois CARDOSO ENST email: cardoso at sig.enst.fr America Liaison Amir ASSADI University of Wisconsin at Madison email: ahassadi at facstaff.wisc.edu Asia Liaison Andrew BACK RIKEN email: andrew.back at usa.net Program Committee Amir Assadi Yianni Attikiouzel John Asenstorfer Andrew Back Geoff Barton Herve Bourlard Andy Chalmers Zheru Chi Andrzej Cichocki Tharam Dillon Tom Downs Hsin Chia Fu Suresh Hangenahally Marwan Jabri Haosong Kong Shigeru Katagiri Anthony Kuh Yi Liu Fa-Long Luo David Miller Christophe Molina M Mohammadian Erkki Oja Soo-Chang Pei Jose Principe Ponnuthurai Suganthan Ah Chung Tsoi Marc Van Hulle A.N. Venetsanopoulos Yue Wang From m.niranjan at dcs.shef.ac.uk Fri Dec 3 03:34:08 1999 From: m.niranjan at dcs.shef.ac.uk (Mahesan Niranjan) Date: Fri, 3 Dec 1999 08:34:08 +0000 (GMT) Subject: Faculty Openings at Sheffield Message-ID: <199912030834.IAA00602@bayes.dcs.shef.ac.uk> Dear Connectionists, A number of faculty positions are being currently advertised in the Department of Computer Science in Sheffield. For one of these, we are looking for someone in the area of Machine Learning - this post was created with my appointment here, and the target candidate will be someone with research interest in Probabilistic Modelling with a signal processing bias. UK academic jobs pay low and are increasingly stressful due to falling Maths standards of students coming in. But this is still a nice job! Please visit http://www.dcs.shef.ac.uk if you might be interested, or pass this to anyone you think might be interested. Closing date 20 December 1999. Many thanks niranjan ____________________________________________________________________ Mahesan Niranjan Phone: 44 114 222 1805 Professor of Computer Science FaX: 44 114 222 1810 The University of Sheffield Email: M.Niranjan at dcs.shef.ac.uk http://www.dcs.shef.ac.uk/~niranjan ____________________________________________________________________ From mdorigo at ulb.ac.be Mon Dec 6 12:17:48 1999 From: mdorigo at ulb.ac.be (Marco DORIGO) Date: Mon, 6 Dec 1999 18:17:48 +0100 Subject: ANTS'2000: Call for papers Message-ID: ANTS'2000 - From Ant Colonies to Artificial Ants: Second International Workshop on Ant Algorithms Brussels, Belgium, September 8-9, 2000 CALL FOR EXTENDED ABSTRACTS (up-to-date information on the workshop is maintained at http://iridia.ulb.ac.be/~ants/ants2000/) SCOPE OF THE WORKSHOP The behavior of social insects in general, and of ant colonies in particular, has since long time fascinated researchers in ethology and animal behavior, who have proposed many models to explain their capabilities. Recently, ant algorithms have been proposed as a novel computational model that replaces the traditional emphasis on control, preprogramming, and centralization with designs featuring autonomy, emergence, and distributed functioning. These designs are proving flexible and robust, able to adapt quickly to changing environments and to continue functioning even when individual elements fail. A particularly successful research direction in ant algorithms, known as "ant colony optimization", is dedicated to their application to discrete optimization problems. Ant colony optimization has been applied successfully to a large number of difficult combinatorial problems like the quadratic assignment and the traveling salesman problems, to routing in telecommunications networks, to scheduling problems, and so on. ANTS'2000 is the second edition of the only event entirely devoted to ant algorithms and to ant colony optimization. Also of great interest to the workshop are models of ant colonies behavior which could stimulate new algorithmic approaches. The aims are to give researchers in both real ants behavior and in ant colony optimization an opportunity to meet, to present their latest research, and to discuss current developments and applications. The two-day event will be held in Brussels, Belgium, from September 8 to September 9, 2000. In the evening of September 7 there will be a tutorial on ant algorithms. RELEVANT RESEARCH AREAS Submissions are invited that describe: (1) Models of aspects of real ant colonies behavior that can stimulate new algorithmic approaches, (2) Empirical and theoretical research in ant algorithms and ant colony optimization. (3) Application of ant algorithms and ant colony optimization methods to real-world problems. SUBMISSION OF PAPERS Papers can be submitted either as extended abstracts or as full papers. The length should be of approximately 4 or 10 pages respectively. Formatting and submission instructions are provided at http://iridia.ulb.ac.be/~ants/ants2000/ants2000-submission.html. Accepted papers will be published in a booklet that is to be distributed to workshop participants. We are now investigating the possibility of publishing the best papers as a journal special issue. In this case, a further call for full papers, that may be longer than the 10 pages workshop format, will be released. All accepted papers will be presented at the workshop as oral presentations or as a poster. REGISTRATION AND FURTHER INFORMATION Up-to-date information about the workshop will be made available at the ANTS'2000 web site (http://iridia.ulb.ac.be/~ants/ants2000/). For information about local arrangements, registration forms, etc., please refer to the above mentioned web site, or contact the local organizer at the address below. A registration fee of BF 7,000 will cover local organization expenses, coffee breaks, and a social dinner on Friday 8 evening. PhD students reduced registration fee: BF 6,000. A proof of inscription is required in that case. IMPORTANT DATES Submission deadline April 15, 2000 Notification of acceptance May 30, 2000 Camera ready copy June 30, 2000 Conference September 8-9, 2000 ANTS'98 CONFERENCE COMMITTEE PROGRAM CHAIR AND LOCAL ARRANGEMENTS Marco DORIGO, IRIDIA, Universite' Libre de Bruxelles, Belgium PROGRAM COMMITTEE Emile AARTS, Philips Research Labs, Eindhoven, The Netherlands Hugues BERSINI, IRIDIA, Universite' Libre de Bruxelles, Belgium Eric BONABEAU, EuroBios, Paris, France Freddy BRUCKSTEIN, Technion, Haifa, Israel Alberto COLORNI, Politecnico di Milano, Italy David CORNE, University of Reading, Great Britain Jean-Louis DENEUBOURG, Universite' Libre de Bruxelles, Belgium Gianni DI CARO, ATR, Kyoto, Japan Marco DORIGO, IRIDIA, Universite' Libre de Bruxelles, Belgium David FOGEL, Natural Selection, La Jolla, CA, USA Luca M. GAMBARDELLA, IDSIA, Lugano, Switzerland Michel GENDREAU, Universite' de Montreal, Canada Deborah GORDON, Stanford University, CA, USA Walter GUTJAHR, University of Vienna, Austria Owen HOLLAND, University of West England, Bristol, UK Vittorio MANIEZZO, Universita' di Bologna, Italy Zbigniew MICHALEWICZ, University of North Carolina, USA Martin MIDDENDORF, Universitaet Karlsruhe, Germany Thomas STUETZLE, Technische Universitaet Darmstadt, Germany Eric TAILLARD, IDSIA, Lugano, Switzerland Guy THERAULAZ, Universite' Paul Sabatier, Toulouse, France Gilles VENTURINI, Universite' de Tours, France SUBMISSIONS ADDRESS Submitted papers should be emailed to: mmi at aifb.uni-karlsruhe.de and to ants at iridia.ulb.ac.be PROGRAM CHAIR ADDRESS Marco Dorigo, Ph.D. Chercheur Qualifie' du FNRS Tel +32-2-6503169 IRIDIA CP 194/6 Fax +32-2-6502715 Universite' Libre de Bruxelles Secretary +32-2-6502729 Avenue Franklin Roosevelt 50 http://iridia.ulb.ac.be/dorigo/dorigo.html 1050 Bruxelles http://iridia.ulb.ac.be/~ants/ants2000/ Belgium http://iridia.ulb.ac.be/dorigo/ACO/ACO.html CONFERENCE LOCATION Avenue A. Buyl 87 Building C - 4th floor (IRIDIA, Dorigo's lab, is at the 5th floor of the same building). There will be arrows and indications to the workshop room. LIMITED NUMBER OF PLACES The number of participants will be limited. If you intend to participate please fill in and send the intention form available at the workshop web page, or drop a line to ants at iridia.ulb.ac.be. RELATED CONFERENCES Note that just after ANTS'2000 there will be in Paris (85 minutes by train from Brussels) two related conferences: SAB2000, September 11-15, and PPSN-2000, September 16-20. From K.Althoefer at kcl.ac.uk Mon Dec 6 13:45:24 1999 From: K.Althoefer at kcl.ac.uk (Althoefer, Kaspar) Date: Mon, 06 Dec 1999 18:45:24 +0000 Subject: Research Studentship in the Dept. of Mechanical Engineering, King's College London References: <36FB78FF.BEE3684A@kcl.ac.uk> Message-ID: <384C0444.73B9F919@kcl.ac.uk> Dear Colleagues, I have a Research Studentship available here in London, UK, to work on motion planning and control algorithms for robots involving research on sensors and neural networks. Applicants should hold a first class honours degree in mechanical engineering, electronic engineering or a related subject. The studentship is only open for candidates from the European Union (EU). Further details can be obtained by writing or e-mailing to me. If you know of anyone interested please feel free to pass this information on. Best regards, Kaspar Althoefer. -- |_/ I N G'S Dr Kaspar ALTHOEFER | \ COLLEGE Ph.D., Dipl.-Ing., AMIEE L O N D O N Department of Mechanical Engineering Founded1829 King's College, Strand, London WC2R 2LS, UK TEL: +44 (0)20 7848 2431, FAX: +44 (0)20 7848 2932 e-mail: K.Althoefer at kcl.ac.uk http://www.eee.kcl.ac.uk/~kaspar From mozer at cs.colorado.edu Tue Dec 7 13:31:21 1999 From: mozer at cs.colorado.edu (Mike Mozer) Date: Tue, 07 Dec 99 11:31:21 -0700 Subject: Positions in machine learning, statistics, and data mining Message-ID: <199912071831.LAA11256@neuron.cs.colorado.edu> Athene Software, Inc. Positions in Machine Learning, Statistics, and Data Mining Athene Software, based in Boulder, Colorado, has immediate openings for professionals in machine learning, statistics, and data mining. We are seeking qualified candidates to develop and enhance models of subscriber behavior for telecommunications companies. Responsibilities include: statistical investigation of large data sets, building predictive and decision-making models using the latest advances in machine learning techniques, developing and tuning data representations, and presentation of results to internal and external customers. Candidates must hold a Ph.D. in Computer Science, Statistics, Electrical Engineering, or related field. The ideal candidate will have experience in pattern recognition or mathematical modeling on real world problems, familiarity with experimental design and data analysis, and some background in relational database systems. Strong communication skills are extremely important. Athene has a long-term committment to cultivating a dynamic, stimulating environment for its Ph.D. research staff. The group is slated to double over the next few years. Athene encourages publication of research results and active participation in the research community. And Athene has established a research advisory board consisting of leaders in machine learning, including Dr. Satinder Singh Baveja (AT&T Labs - Research), Prof. Geoffrey Hinton (University College London), Prof. John Moody (OGI), Prof. Andrew Moore (CMU), and Prof. Michael Mozer (Boulder). Send applications to: Dr. Robert Dodier Athene Software, Inc. 2060 Broadway, Suite 300 Boulder, CO 80302 email: robertd at athenesoft.com company URL: www.athenesoft.com From AJain at cc.ucsf.edu Tue Dec 7 13:55:09 1999 From: AJain at cc.ucsf.edu (Jain, Ajay) Date: Tue, 7 Dec 1999 10:55:09 -0800 Subject: Research Positions at UCSF: Computational biology, bioinformatics, databases Message-ID: <112CB72C222BD2118FFF00A0C9D18C6460AEF6@cc.ucsf.edu> ABOUT THE POSITIONS: We are seeking two individuals for post-doctoral positions in computational research applied to the study of cancer. Candidates for these positions must have a PhD in computer science or closely related field and they ideally will have experience applying sophisticated computation to real-world problems (e.g. computational biology, drug discovery, object recognition, robotics, etc.). Experience in machine-learning, neural networks, computational geometry, or physical modeling is beneficial, as is formal training in chemistry, biology, or physics. Experience in applying machine learning techniques to problems involving noisy data is particularly relevant. Candidates will be expected to learn aspects of disciplines complementary to their own and collaborate extensively with clinical and biological researchers. We are also seeking an individual for a challenging database development position. This position is responsible for design and implementation of database and data processing systems to support clinical and bench science efforts in a web-based environment. Designing databases to track human tissue samples, databases for long-term patient follow-up, and deploying nationally accessible biological data repositories all fall under the responsibilities of this position. The systems must be optimized for real-time content retrieval via user-friendly Web interfaces. This position is for a senior developer able to take on a leadership role, supervising and mentoring other developers. Candidates for all of these positions must have excellent communication skills, excellent work history and references, and the ability to work both independently and as part of a multidisciplinary team of scientists, including colleagues from biology, medicine, chemistry, genetics, and other fields. We offer competitive salaries and benefits. UC is an affirmative action and equal opportunity employer. ABOUT THE PLACE (see http://cc.ucsf.edu): Housed within one of the nation's top biomedical research universities, the UCSF Cancer Center brings together hundreds of researchers and clinicians dedicated to four fundamental pursuits: basic science research into the causes and events of cancer's progression; clinical research to translate new knowledge into viable treatments; sensitive, state-of-the-art patient care; and programs in epidemiology, prevention, and cancer control. Basic scientists and clinical health care providers work together toward the translation of promising ideas into the creation of new and better treatments for cancer patients everywhere. ABOUT THE RESEARCH (see http://cc.ucsf.edu/jain): Cancer is a disease marked by accumulating genetic abnormalities, ultimately resulting in malignancy. These abnormalities are evident in the genomic DNA, messenger RNA, and in the expressed proteins of cancer cells. Modern techniques for in vitro genomic and proteomic analysis are generating data on a very large scale (up to 30,000 data points per experiment). DNA microarrays, comparative genomic hybridization, flourescent in situ hybridization, 2D gel electrophoresis, and other techniques produce quantitative data that contain information relevant to the biological state of cells. These data offer an enormous opportunity for systematic computational analyses to make an impact on both therapeutic target discovery as well as in clinical decision-making. There is a pressing need for computational methods to move beyond data visualization and toward generating conclusions that are quantitatively supportable. Coupled with data emerging from the Human Genome Project and modern structure-based drug design techniques, there are significant opportunities to advance cancer research and treatment. Members of the growing Jain Lab are engaged in developing and applying novel computational techniques for: 1) Analyzing biological data gathered from human and animal tumors generated within the Center and with external collaborators. 2) Relating DNA/protein sequences to protein function. 3) Structure-based small-molecule drug design aimed at discovery or refinement of cancer therapeutics. We are seeking energetic individuals to join in our research efforts. ABOUT DR. AJAY N. JAIN: Dr. Jain received his PhD in 1991 from Carnegie Mellon University working with Dr. Dave Touretzky and Dr. Alex Waibel. From 1992 through mid-1999, he worked in a series of start-up biopharmaceutical companies on computational methods for structure-based drug design. Dr. Jain is Director of Informatics at the UCSF Cancer Center and holds faculty appointments in the Cancer Research Institute and Dept. of Laboratory Medicine (see http://cc.ucsf.edu/people/jain_ajay.html). TO APPLY: Please send a full CV and letter describing your qualifications by Fax, US Mail, or email (pdf, plain text, or MS Word). Email is the preferred mode. Contact information follows. ____________________________________________________ Ajay N. Jain, PhD Director of Informatics, UCSF Cancer Center Box 0128, San Francisco, CA 94143-0128 Deliveries: 2340 Sutter St., Room S336, San Francisco Tel: (415) 502-7242 Fax: (415) 476-8218 Email: ajain at cc.ucsf.edu Web: http://cc.ucsf.edu/jain From m.towsey at qut.edu.au Tue Dec 7 21:52:30 1999 From: m.towsey at qut.edu.au (Michael Towsey) Date: Wed, 08 Dec 1999 12:52:30 +1000 Subject: Special Issue of "Applied Intelligence" Message-ID: <384DC7EE.21BD18AE@fit.qut.edu.au> CALL FOR CONTRIBUTIONS The submission of papers is invited for a special issue on "Neural Networks and Machine Learning for Natural Language Processing" in "Applied Intelligence: The International Journal of Artificial Intelligence, Neural Networks, and Complex Problem-Solving Techniques" GUEST EDITORS: Joachim Diederich Claudia Brugman Michael Towsey ISSUE THEME: At the core of several disciplines is the question of natural language processing, and its relationship, mediated by language acquisition, to innate linguistic knowledge. Studies in neurophysiology, as well as recent connectionist natural language processing models, have challenged the "poverty of the stimulus" argument and the existence of "Universal Grammar" as innate grammatical structure. Classical generative linguistic theory assumes a modular, encapsulated and autonomous representation of grammatical knowledge which is independent of other representations. This assumption is consistent with the view that grammatical knowledge is genetically programmed and pre-wired in the brain and implies at the level of evolution a possible sudden emergence of linguistic competence, a capacity which is not shared by other species. Related to the issues of innateness and acquisition is the issue of language universals. As described above, "Universal Grammar" is characterised in terms of innate and modular structures. The relationship between this notion of universal and the empiricist notion of language universal, in terms of properties all languages share, has also been the object of hypothesis and study, and both must be accounted for in any model of the innate aspects of language competence. Moreover, any theory that addresses the neurobiological basis of language acquisition must address the question of how the brain can realise the computation required for language processing. We invite articles which address any of these related issues from the point of view of neural network representation or more general, machine learning for natural language processing. At the core of the research topics outlined above is the question of how much prior knowledge is necessary for natural language learning and how this question can be addressed by theoretical or empirical studies. The theme of this special issue comprises * the investigation of the underlying theoretical foundations, including computational learning theory * the implementation and evaluation of methods for learning natural language with neural networks or other machine learning approaches, and * applications of such learning systems in various domains. TOPICS OF INTEREST The list below gives some examples of intended topics: * spoken language acquisition * connectionist approaches for learning the syntax and semantics of natural languages * statistical approaches for language acquisition * hybrid systems (symbolic machine learning/neural network) for learning natural languages * genetic algorithms for language acquisition * inductive learning for language acquisition, including decision-tree learners and inductive logic programming. * comparative evaluations of different machine learning methods SUBMISSION PROCESS Prospective authors should send an electronic mail message by 10th January 2000 indicating their intent to submit a paper to the guest editor of the special issue, Prof. Joachim Diederich (joachim at fit.qut.edu.au). This message should contain a preliminary abstract and three to five keywords. Six hard copies of the final manuscript should be sent to the guest editor (not to the Applied Intelligence Editorial office) at the following address: Professor Joachim Diederich Machine Learning Research Centre Faculty of Information Technology Queensland University of Technology GPO 2434 Brisbane 4001 QLD Australia To speed up the reviewing process, authors should also send a PostScript version of the paper via email to the guest editor. Prospective authors can find further information about the journal on the home page http://kapis.www.wkap.nl/journalhome.htm/0924-669X SCHEDULE: Expressions of interest: January 10, 2000 Paper submission deadline: March 31, 2000 Review decision by: May 31, 2000 Final manuscript due: July 31, 2000 _________________________________________________ Submitted by Michael Towsey From yoram at stat.Berkeley.EDU Wed Dec 8 13:26:52 1999 From: yoram at stat.Berkeley.EDU (YORAM Gat) Date: Wed, 8 Dec 1999 10:26:52 -0800 (PST) Subject: Paper on generalization bounds for the support vector method Message-ID: The following paper is now available at http://www.stat.Berkeley.EDU/tech-reports/548.ps.Z A bound concerning the generalization ability of a certain class of learning algorithms Yoram Gat Abstract: A classifier is said to have good generalization ability if it performs on test data almost as well as it does on the training data. The main result of this paper provides a sufficient condition for a learning algorithm to have good finite sample generalization ability. This criterion applies in some cases where the set of all possible classifiers has infinite VC dimension. We apply the result to prove the good generalization ability of support vector machines. From terry at salk.edu Wed Dec 8 23:39:36 1999 From: terry at salk.edu (terry@salk.edu) Date: Wed, 8 Dec 1999 20:39:36 -0800 (PST) Subject: COMPUTATIONAL NEUROBIOLOGY GRADUATE PROGRAM Message-ID: <199912090439.UAA01824@hebb.salk.edu> DEADLINE: JANUARY 7, 2000 COMPUTATIONAL NEUROBIOLOGY GRADUATE PROGRAM Department of Biology -- University of California, San Diego http://www.biology.ucsd.edu/compneuro/ The goal of the Computational Neurobiology Graduate Program at UCSD is to train a new generation of researchers who are equally at home measuring large-scale brain activity, analyzing the data with advanced computational techniques, and developing new models for brain development and function. Candidates from a wide range of backgrounds are invited to apply, including Biology, Psychology, Computer Science, Physics and Mathematics. The three major themes in the training program are: 1. Neurobiology of Neural Systems -- Anatomy, physiology and behavior of systems of neurons. Using modern neuroanatomical, neuropharmacological and electrophysiological techniques. Lectures, wet laboratories and computer simulations, as well as research rotations. Major new imaging and recording techniques also will be taught, including two-photon laser scanning microscopy and functional magnetic resonance imaging (fMRI). 2. Algorithms and Realizations for the Analysis of Neuronal Data -- New algorithms and techniques for analyzing data obtained from physiological recording, with an emphasis on recordings from large populations of neurons with imaging and multielectrode recording techniques. New methods for the study of co-ordinated activity, such as multi-taper spectral analysis and Independent Component Analysis (ICA). 3. Neuroinformatics, Dynamics and Control of Systems of Neurons -- Theoretical aspects of single cell function and emergent properties as many neurons interact among themselves and react to sensory inputs. A synthesis of approaches from mathematics and physical sciences as well as biology will be used to explore the collective properties and nonlinear dynamics of neuronal systems, as well as issues of sensory coding and motor control. Requests for application materials should be sent to the Graduate Admissions Office, Department of Biology 0348, 9500 Gilman Drive, UCSD, La Jolla, CA, 92093-0348: [gradprog at biology.ucsd.edu]. The deadline for completed application materials, including letters of reference, is JANUARY 7, 2000. More information about applying to the UCSD Biology Graduate Program: http://www-biology.ucsd.edu/sa/Admissions.html The Biology Department home page is located at: http://www-biology.ucsd.edu/ Other inquiries about the Computational Neurobiology Graduate Program should be directed to: Terrence Sejnowski Institute for Neural Computation 0523 University of California, San Diego La Jolla, CA 92093 tsejnowski at ucsd.edu Participating Faculty include: * Henry Abarbanel (Physics): Nonlinear and oscillatory dynamics; modeling central pattern generators in the lobster stomatogastric ganglion. Director of the Institute for Nonlinear Systems at UCSD. * Thomas Albright (Salk Institute): Motion processing in primate visual cortex; linking single neurons to perception; fMRI in awake, behaving monkeys. Director, Sloan Center for Theoretical Neurobiology. * Darwin Berg (Biology): Regulation synaptic components, assembly and localization, function and long-term stability. Former Chairman of Biology. * Garrison Cottrell (Computer Science and Engineering): Dynamical neural network models and learning algorithms. * Mark Ellisman (Neurosciences, School of Medicine): High resolution electron and light microscopy; anatomical reconstructions. Director, National Center for Microscopy and Imaging Research. * Robert Hecht-Nielsen (Electrical and Computer Engineering): Neural computation and the functional organization of the cerebral cortex. Founder of Hecht-Nielsen Corporation. * Harvey Karten (Neurosciences, School of Medicine): Anatomical, physiological and computational studies of the retina and optic tectum of birds and squirrels. * David Kleinfeld (Physics):Active sensation in rat somatosensation; properties of neuronal assemblies; optical imaging of large-scale activity. Co-director, Analysis of Neural Data Workshop (MBL). * William Kristan (Biology): Neuroethology of leech; functional and developmental studies of the leech nervous system, including computational studies of the bending reflex and locomotion. Director of the Neurosciences Graduate Program. * Herbert Levine (Physics): Nonlinear dynamics and pattern formation in physical and biological systems, including cardiac dynamics and the growth and form of bacterial colonies. * Javier Movellan (Cognitive Science): Sensory fusion and learning algorithms for continuous stochastic systems. * Mu-ming Poo (Biology): Mechanisms for synaptic plasticity; developmental plasticity and learning in nervous systems; development of sensory maps in lower vertebrate visual systems. * Mikhael Rabinovich (Institute for Nonlinear Science): Dynamical systems analysis of the stomatogastric ganglion of the lobster and the antenna lobe of insects. * Terrence Sejnowski (Salk Institute/Biology): Computational neurobiology; physiological studies of neuronal reliability and synaptic mechanisms. Director, Institute for Neural Computation. * Martin Sereno (Cognitive Science): Neural bases of visual cognition and language using anatomical, electrophysiological, computational, and non-invasive brain imaging techniques. * Nicholas Spitzer (Biology): Regulation of ionic channels and neurotransmitters in neurons; effects of electrical activity in developing neurons on neural function. Chair of the Neurobiology Section. * Charles Stevens (Salk Institute): Synaptic physiology; physiological studies and biophysical models of synaptic plasticity in hippocampal neurons. * Roger Tsien (Chemistry): Second messenger systems in neurons; development of new optical and MRI probes of neuron function, including calcium indicators and caged neurotransmitters. * Mark Whitehead (Neurosurgery, School of Medicine): Peripheral and central taste systems; anatomical and functional studies of regions in the caudal brainstem important for feeding behavior. * Ruth Williams (Mathematics): Probabilistic analysis of stochastic systems and continuous learning algorithms. From Volker.Tresp at mchp.siemens.de Wed Dec 8 09:29:45 1999 From: Volker.Tresp at mchp.siemens.de (Volker Tresp) Date: Wed, 08 Dec 1999 15:29:45 +0100 Subject: Paper available on combining estimators and Gaussian process regression Message-ID: <384E6B59.DBB38C36@mchp.siemens.de> I would like to announce the availability of a new paper of potential interest to people working on combining estimators and Gaussian process regression or other kernel based systems. Comments are welcome! Greetings, - Volker ------------------------------------------------------------------------- A BAYESIAN COMMITTEE MACHINE Volker Tresp Siemens AG, Corporate Technology Neural Computation Dept. Information and Communications Otto-Hahn-Ring 6, 81730 Munich, Germany The Bayesian committee machine (BCM) is a novel approach to combining estimators which were trained on different data sets. Although the BCM can be applied to the combination of any kind of estimators the main foci are Gaussian process regression and related systems such as regularization networks and smoothing splines for which the degrees of freedom increase with the number of training data. Somewhat surprisingly, we find that the performance of the BCM improves if several test points are queried at the same time and is optimal if the number of test points is at least as large as the degrees of freedom of the estimator. The BCM also provides a new solution for online learning with potential applications to data mining. We apply the BCM to systems with fixed basis functions and discuss its relationship to Gaussian process regression. Finally, we also show how the ideas behind the BCM can be applied in a frequentist setting to extend the input dependent combination of estimators. ftp://flop.informatik.tu-muenchen.de/pub/hofmannr/bcm.ps.gz ----------------------------------------------------------------------- ----------------------------------------------------------------------- From sylee at ee.kaist.ac.kr Thu Dec 9 03:49:55 1999 From: sylee at ee.kaist.ac.kr (Soo-Young Lee) Date: Thu, 9 Dec 1999 17:49:55 +0900 Subject: Position Available : Research Professor and Post Doc Message-ID: <003d01bf4222$63b21640$329ef88f@kaist.ac.kr> I apologize for multiple receipants. The Brain Science Research Center (BSRC) and Department of Electrical and Computer Science at Korea Advanced Institute of Science and Technology have immediate openings for Research Professor and Post Doc. The Research Professor position requires at least 1 year experience after Ph.D. degree. The BSRC is the main research organization for Korean Brain Science and Engineering Research Program sponsored by Korean Ministry of Science and Technology. It currently manages a big national research program, and more than 100 professors throughout Korea are working in the research program. It is truly multidisciplinary research program, and ample opertunities exist to cooparate with researchers in other disciplines, from neuroscisnce to neural network applications and implementations. Although all aspects of biological and artificial neural neural networks are pursued, researchers in the following subjects have higher priority. (1) cognitive science (selective attention, etc.) (2) speech signal processing and recognition (3) auditory model (4) neuro-chip implementation (5) dynamic neural models WE EMPHASIZE INTERDISCIPLINARY RESEARCHES, FROM BIOLOGY TO HARDWARE THROUGH MODELING AND APPLICATIONS. Interested candidates should contact with Prof. Soo-Young Lee, Director, BSRC, at sylee at ee.kaist.ac.kr. From Sebastian_Thrun at heaven.learning.cs.cmu.edu Thu Dec 9 19:32:35 1999 From: Sebastian_Thrun at heaven.learning.cs.cmu.edu (Sebastian Thrun) Date: Thu, 09 Dec 1999 19:32:35 -0500 Subject: Call for papers: ICML-2k Message-ID: Please find enclosed the CFP for ICML-2000. It might interest you that ICML seeks to broaden its scope, and among other areas solicits submissions from the connectionists/statistical learning community. Consider submitting! It's a fun conference. -------------------------------------------------------------------------- Call for Papers THE SEVENTEENTH INTERNATIONAL CONFERENCE ON MACHINE LEARNING June 29-July 2, 2000 Stanford University The Seventeenth International Conference on Machine Learning (ICML-2000) will be held at Stanford University from June 29 to July 2, 2000, in the heart of Silicon Valley. The conference will bring together researchers to exchange ideas and report recent progress in the computational study of learning. Topics for Submission ICML-2000 welcomes submissions on all facets of machine learning, but especially solicits papers on problem areas, research topics, learning paradigms, and approaches to evaluation that have been rare at recent conferences, including: - the role of learning in natural language, vision and speech, planning and scheduling, design and configuration, logical and spatial reasoning, motor control, and more generally on learning for performance tasks carried out by intelligent agents; - the discovery of scientific laws and taxonomies, the construction of componential and structural models, and learning at multiple levels of temporal and spatial resolution; - the effect of the developers' decisions about problem formulation, representation, data quality, and reward function on the learning process; - computational models of human learning, applications to real-world problems, exploratory research that describes novel learning tasks, work that integrates familiar methods to demonstrate new functionality, and agent architectures in which learning plays a central role; - empirical studies that combine natural data (to show relevance) with synthetic data (to understand conditions on behavior), along with formal analyses that make contact with empirical results, especially where the aim is to identify sources of power, rather than to show one method is superior to others. Naturally, we also welcome submissions on traditional topics, ranging from induction over supervised data to learning from delayed rewards, but we hope the conference will also attract contributions on the issues above. Review Process The ICML-2000 review process will be structured to encourage publications covering a broad range of research and to foster increased participation in the conference. To this end, we have instituted: - area chairs who will be responsible for recruiting papers in their area of expertise and overseeing the review process for those submissions; - conditional acceptance of papers that are not publishable in their initial form, but that can be improved enough for inclusion in time to appear in the proceedings; and - a review form that requires referees to explicitly list any problems with a paper, what it would take to overcome them, and, if they recommend rejection, why it cannot be fixed in time for inclusion. The overall goal is to make the review process more like that in journals, with time for the authors to incorporate feedback from reviewers. Each submitted paper will be reviewed by two members of the program committee, with the decision about its acceptance overseen by the responsible area chair and the program chair. Paper Submission Authors should submit papers using same format and length as the final proceedings version. The detailed instructions for authors at http://www-csli.stanford.edu/icml2k/instructions.html include pointers to templates for LaTeX and Word documents. These specify two-column style, Times Roman font with 10 point type, vertical spacing of 11 points, overall text width of 6.75 inches, length of 9.0 inches, 0.25 inches between the two columns, top margin of 1.0 inch, and left margin of 0.75 inch. (The right and bottom margins will depend on whether one uses US letter or A4 paper.) Papers must not exceed eight (8) pages including figures and references. We will return to the authors any papers that do not satisfy these requirements. The deadline for submissions to ICML-2000 is MONDAY, JANUARY 24, 2000. Submission will be entirely electronic by transferring papers to the ICML-2000 ftp site, as explained in the detailed instructions for authors. Authors must submit papers in POSTSCRIPT format to ensure our ability to print them out for review. Each submission must be accompanied by the paper's title, the authors' names and physical addresses, a 250-word abstract, the contact author's email address and phone number, and the author who would present the talk at the conference. Authors must enter this information into the submission form at the conference web site by FRIDAY, JANUARY 21. ICML-2000 allows simultaneous submission to other conferences, provided this fact is clearly indicated on the submission form. Accepted papers will appear in the conference proceedings only if they are withdrawn from other conferences. Simultaneous submissions that are not clearly specified as such will be rejected. Other Conference Information The Seventeenth International Conference on Machine Learning will be collocated with the Thirteenth Annual Conference on Computational Learning Theory (COLT-2000) and the Sixteenth Conference on Uncertainty in Artificial Intelligence (UAI-2000). Registrants to any of these meetings will be able to attend the technical sessions of the others at no additional cost. ICML-2000 will also be preceded by tutorials on various facets of machine learning. For additional information, see the web site for the conference at http://www-csli.stanford.edu/icml2k/ which will provide additional details as they become available. If you have questions about ICML-2000, please send electronic mail to icml2k at csli.stanford.edu. The conference has received support from DaimlerChrysler Research and Technology, Stanford's Center for the Study of Language and Information (CSLI), and the Institute for the Study of Learning and Expertise (ISLE). From plong at comp.nus.edu.sg Thu Dec 9 23:11:02 1999 From: plong at comp.nus.edu.sg (Phil Long) Date: Fri, 10 Dec 1999 12:11:02 +0800 (GMT-8) Subject: computational learning theory postdoc Message-ID: The computational learning theory group of the National University of Singapore is looking for a postdoc. The focus of the project funding the position is the search for theoretically principled, practical algorithms for machine learning. The postdoc will be free to pursue independent research along these general lines, but will also have the opportunity to collaborate with the members of the group. The starting date is somewhat flexible, but should be some time before September, 2000. The position runs for two years. If you are interested, please send your CV and the names of three references to plong at comp.nus.edu.sg by January 15, 2000. From urbanczr at aston.ac.uk Fri Dec 10 09:18:55 1999 From: urbanczr at aston.ac.uk (R URBANCZIK) Date: Fri, 10 Dec 1999 14:18:55 +0000 Subject: paper on Imax Message-ID: The following paper (12 pages, submitted to Europhysics letters) is available online from: http://neural-server.aston.ac.uk/Papers/postscript/NCRG_1999_033.ps.Z Statistical Mechanics of Mutual Information Maximization R. Urbanczik, Neural Computing Research Group, Aston University Abstract: An unsupervised learning procedure based on maximizing the mutual information between the outputs of two networks receiving different but statistically dependent inputs is analyzed (Becker and Hinton, Nature, 355, 92, 161). By exploiting a formal analogy to supervised learning in parity machines, the theory of zero temperature Gibbs learning for the unsupervised procedure is presented for the case that the networks are perceptrons and for the case of fully connected committees. From rajeev at cns.bu.edu Fri Dec 10 18:22:19 1999 From: rajeev at cns.bu.edu (Rajeev Raizada) Date: Fri, 10 Dec 1999 18:22:19 -0500 (EST) Subject: Review of MIT CogSci Encyc downloadable Message-ID: Members of Connectionists might be interested in the following book review of "The MIT Encyclopedia of the Cognitive Sciences", forthcoming in the journal Neural Networks. A preprint is downloadable in PDF or PostScript formats at: http://cns-web.bu.edu/~rajeev/index.html#publications R. D. S. Raizada (1999). "A fruitful blend, or a trinket-box? A book review of The MIT Encyclopedia of the Cognitive Sciences." Forthcoming in Neural Networks. This review compares the MIT Encyclopedia to other recent reference works covering similar subject matter, e.g. M. S. Gazzaniga (Ed.) (1999) "The New Cognitive Neurosciences - 2nd Edition", and also discusses the extent to which it succeeds in showing interdisciplinary links between the various branches of Cognitive Science. Please let me know if you have any problems with downloading the preprint. Yours, Rajeev Raizada Dept. of Cognitive and Neural Systems Boston University 677 Beacon St. Boston, MA 02215. Tel: 617-353-6181 Fax: 617-353-7755 E.mail: rajeev at cns.bu.edu WWW: http://cns-web.bu.edu/~rajeev From F.Murtagh at Queens-Belfast.ac.uk Sat Dec 11 15:05:20 1999 From: F.Murtagh at Queens-Belfast.ac.uk (Prof. F. Murtagh) Date: Sat, 11 Dec 1999 20:05:20 +0000 (GMT) Subject: clustering and classification - _the_ source for bibliographies Message-ID: Dear Colleague, The following may be of interest, either to contribute a software or other exhibit area, or to get a copy when available (approx. May 2000). Regards, ------------------------------------------------------------- Prof. F. Murtagh, School of Computer Science, The Queen's University of Belfast, Belfast BT7 1NN, Northern Ireland Tel +44 28 9027-4620 Fax +44 28 9068-3890 http://www.cs.qub.ac.uk/~F.Murtagh f.murtagh at qub.ac.uk Centre for Image and Vision Systems http://www.qub.ac.uk/ivs ------------------------------------------------------------- Clustering and Classification Bibliographies - and now Software The Classification Literature Automated Search Service (CLASS), the only classification bibliography, published annually since the 1970s by the Classification Society of North America (CSNA), is undergoing big changes. From susanne at smi.auc.dk Mon Dec 13 04:48:30 1999 From: susanne at smi.auc.dk (Susanne Nielsen) Date: Mon, 13 Dec 1999 10:48:30 +0100 Subject: NEW DEADLINE FOR REGISTRATION AND SUBMISSION OF ABSTRACTS FOR IFESS 2000, NP 2000 WORKSHOP, AND PRE-CONFERENCE COURSE Message-ID: IFESS - YEAR 2000 - AALBORG, DENMARK ------------------------------------ IFESS 2000 CONFERENCE, 18TH-20TH JUNE NP 2000 WORKSHOP, 21ST-23RD JUNE PRE-CONFERENCE COURSE - MOTOR FUNCTIONS AND CNS INJURIES, 17TH-18TH JUNE NB: BECAUSE OF THE CHRISTMAS HOLIDAY WE HAVE POSTPONED THE DEADLINE FOR TWO WEEKS. NEW DEADLINE FOR REGISTRATION AND SUBMISSION OF ABSTRACTS IS THEREFORE 7TH JANUARY 2000 (POSTMARKED). http://www.smi.auc.dk/aalborg2000 1. The pre-conference course: "Motor Functions and CNS Injuries" - is designated as an international workshop to facilitate the communication between the various disciplines involved in the field of restoration of motor and sensory functions of individuals with motor disabilities - 17th-18th June 2000. See: http://www.smi.auc.dk/aalborg2000/course 2. The 5th Annual Conference of the International Functional Electrical Stimulation Society (IFESS 2000) - aims at bringing together internationally leading specialists in the field of Functional Electrical Stimulation (FES) to stimulate the exchange of views and new advancements and to discuss current and future goals in various applications of FES - 18th-20th June 2000. See: http://www.smi.auc.dk/aalborg2000/ifess 3. The Neural Prostheses VI: Motor Systems Conference (NP 2000) - will be organized as an international workshop to facilitate the communication between the various disciplines involved in the field of restoration of motor and sensory functions of individuals with motor disabilities - 21st-23th June 2000. See: http://www.smi.auc.dk/aalborg2000/np It is the first time in Europe that these events are organized simultaneously. The organizing Committee has aimed at having distinguished keynote and invited speakers to introduce all main topics, workshops, and sessions. We are expecting to have up to 200 participants from all over the world, among them about 50% graduate students and many clinicians. As a result of funding of the IFESS 2000, we now have grants available for Young Researchers. Further information can be found at: http://www.smi.auc.dk/aalborg2000/grants/grants.html The scientific programs for all three events, information on practical details, and registration form can be found at: http://www.smi.auc.dk/aalborg2000. If you have any further questions, please do not hesitate contacting us at: aalborg2000 at smi.auc.dk or at the following address: Center for Sensory-Motor Interaction (SMI) Aalborg University Fredrik Bajers Vej 7D-3 9220 Aalborg Denmark Phone: +45 96 35 88 28 Fax: +45 98 15 40 08 I take this oppotunity to wish you a merry Christmas and a happy New Year and do look forward to seeing you in Aalborg in the new millenium. Best regards, Thomas Sinkjaer Chairman Organising Committee IFESS 2000 & NP 2999 Professor, dr.med., Ph.D. --------------------------******--------------------------- Susanne Nielsen, Research Secretary Center for Sensory-Motor Interaction (SMI) Aalborg University Fredrik Bajers Vej 7D-3 9220 Aalborg Denmark Phone: +45 96 35 88 28 Fax: +45 98 15 40 08 E-mail: susanne at smi.auc.dk --------------------------******--------------------------- From psollich at mth.kcl.ac.uk Thu Dec 16 07:37:50 1999 From: psollich at mth.kcl.ac.uk (Peter Sollich) Date: Thu, 16 Dec 1999 12:37:50 +0000 (GMT) Subject: Postgraduate research open day Message-ID: Dear colleagues, on behalf of the Neural Networks and Disordered Systems group in the Department of Mathematics at King's College London, I am forwarding the announcement below of our forthcoming postgraduate open day. I would be grateful if you could encourage all interested (and suitably qualified) students to attend. Many thanks in advance for your help, Peter Sollich -------------------------------------------------------------------------- Peter Sollich Department of Mathematics Phone: +44 - (0)20 - 7848 2875 (new!) King's College Fax: +44 - (0)20 - 7848 2017 (new!) University of London E-mail: peter.sollich at kcl.ac.uk Strand WWW: http://www.mth.kcl.ac.uk/~psollich London WC2R 2LS, U.K. -------------------------------------------------------------------------- ***************************************************************************** * * * POSTGRADUATE OPEN DAY * * --------------------- * * * * The Mathematics Department of King's College London is holding an Open * * Day for prospective postgraduate students on the 18th of February, * * 2000. The day will include talks by members of the various research * * groups in the Department, as well as a panel discussion and small group * * discussions with prospective supervisors and current students. Coffee, * * lunch and tea will be provided, and the department will consider * * making a contribution towards students' travel expenses. * * * * King's College has one of the largest departments of mathematics in * * the United Kingdom and is one of the four London Colleges with the * * highest RAE research rating in both pure and applied mathematics. The * * mathematics department of King's is an exceptionally active, balanced * * and broad research unit, which offers research supervision in a wide * * range of subjects in pure and applied mathematics, including Analysis, * * Number Theory, Geometry and Topology, Theoretical Physics, General * * Relativity, * * NEURAL NETWORKS AND DISORDERED SYSTEMS, * * * * Statistical Mechanics, Quantum Probability, Mathematical Biology and * * Financial Mathematics. * * * * For full details of the Open Day programme, a registration form, and * * the booklet "Postgraduate Mathematics at King's", please contact * * * * The Postgraduate Secretary, Department of Mathematics, * * King's College London, Strand, London WC2R 2LS * * Phone 020-7848 2107, Fax 020-7848 2017, Email pgopenday.maths at kcl.ac.uk * * * * Further details and an electronic registration form are also available * * from the Department's web site at * * "http://www.mth.kcl.ac.uk/postgraduate/openday2000.html". * * * ***************************************************************************** From Roweis at gatsby.ucl.ac.uk Thu Dec 16 08:02:39 1999 From: Roweis at gatsby.ucl.ac.uk (Sam T. Roweis) Date: Thu, 16 Dec 1999 13:02:39 +0000 (GMT) Subject: NIPS tutorial notes on the web Message-ID: NIPS'99 Tutorial Notes Probabilistic Models for Unsupervised Learning Zoubin Ghahramani Sam Roweis We are making the slides from our recent tutorial available for download on the web. These slides correspond to the ones handed out in the NIPS tutorial booklet and also to those we used during the presentation. Feel free to distribute them as a package, but please let us know if you plan to include parts them in other documents. http://www.gatsby.ucl.ac.uk/~zoubin/NIPStutorial.html Enjoy the holidays, Sam & Zoubin ____________________________________________________________________________ Sam Roweis Gatsby Computational Neuroscience Unit, UCL 44 20 7 679 1188 voice 17 Queen Square, Room 503 44 20 7 679 1173 fax WC1N 3AR, London, UK roweis at gatsby.ucl.ac.uk http://www.gatsby.ucl.ac.uk/~roweis/ From eurich at theo.physik.uni-bremen.de Fri Dec 17 04:37:00 1999 From: eurich at theo.physik.uni-bremen.de (Christian Eurich) Date: Fri, 17 Dec 1999 10:37:00 +0100 Subject: Article on Hebbian learning Message-ID: <385A043C.93EB7A0C@physik.uni-bremen.de> Dear Connectionists, on the occasion of the recent NIPS workshop "Spike Timing and Synaptic Plasticity" organized by Paul Munro and Larry Abbott I would like to call your attention to the following article on the problem of delay adaptation and the stability of time sensitive Hebbian learning for different window functions. The article appeared in the Feb 15 issue of Physical Review Letters. Hardcopies are available on request. Christian W. Eurich, Klaus Pawelzik, Udo Ernst, Jack D. Cowan and John G. Milton, Dynamics of Self-Organized Delay Adaptation, Physical Review Letters 82 (1999) 1594-1597. Abstract: Adaptation of interaction delays is essential for the functioning of many natural and technical systems. We introduce a novel framework for studying the dynamics of delay adaptation in systems which optimize coincidence of inputs. For the important case of periodically modulated input we derive conditions for the existence and stability of solutions which constrain the set of mechanisms for reliable delay adaptation. Using numerical examples we show that our approach is applicable to more general than periodic input patterns such as Poissonian point processes with coordinated rate fluctuations. Cheers, Christian Eurich ================================================== Dr. Christian Eurich Universitaet Bremen Institut fuer Theoretische Neurophysik, FB 1 Postfach 330 440 D-28334 Bremen Germany Phone: +49 (421) 218-4559 Fax: +49 (421) 218-9104 Phone home: +49 (421) 5577208 Fax home: +49 (421) 5577050 e-mail: eurich at physik.uni-bremen.de homepage: http://www-neuro.physik.uni-bremen.de/~eurich "Die Welt ist alles, was der Fall ist." (L. W.) ================================================== From p.j.b.hancock at psych.stir.ac.uk Fri Dec 17 07:35:10 1999 From: p.j.b.hancock at psych.stir.ac.uk (Peter Hancock) Date: Fri, 17 Dec 1999 12:35:10 +0000 (GMT) Subject: Faculty posts in Computing Science Message-ID: The computing science department at the University of Stirling has vacancies for two chairs (full professors) and two lecturers. They would be very interested in people doing research in neural networks, to join the existing interdisciplinary group here. The official closing date has past, but anyone interested might still contact Prof. Ken Turner, kjt at cs.stir.ac.uk. http://www.cs.stir.ac.uk/jobs/ ------------------------------------------------------------------------- Peter Hancock Department of Psychology 0 0 Face Centre for University of Stirling | Research and Cognitive and FK9 4LA, UK \_/ Group Computational Phone 01786 467675 Fax 01786 467641 Neuroscience pjh at psych.stir.ac.uk http://www-psych.stir.ac.uk/~pjh ------------------------------------------------------------------------- From stefano.panzeri at ncl.ac.uk Fri Dec 17 09:23:19 1999 From: stefano.panzeri at ncl.ac.uk (Stefano Panzeri) Date: Fri, 17 Dec 1999 14:23:19 +0000 Subject: postdoctoral position in computational vision Message-ID: <385A4757.333556D8@ncl.ac.uk> Post-doctoral position in computational vision at the Neural Systems Group, Department of Psychology, University of Newcastle upon Tyne, UK http://www.psychology.ncl.ac.uk/neural_systems_group.html Applications are invited for a post-doctoral scientist position (up to ?30,626, depending on experience) for a research in computational neuroscience. The post is funded for 3 years in a first instance by an EU programme aimed at using concurrent psychophysics, computational models and neurophysiology to investigate mechanisms in visual neurons during natural image processing. Research topics for the computational neuroscience position will include: computational models of properties of visual neurons when they are stimulated with natural images; computational models of the role of prior knowledge of the statistics of the natural visual environment in visual perception; quantitative analysis of the response properties of visual neurons responding to natural images; developing new computational approaches for data in structure-function relationships (in particular relating connectivity, activation and information dynamics in the thalamo-cortical network). Applicants should possess a PhD in a relevant discipline (i.e. neuroscience, physics, computing, applied mathematics, engineering or others), a strong analytical background and a keen interest in vision. Experience and skills either in neural networks, information theory, visual psychophysics, visual neuroscience or computer image processing will be a great advantage. Departmental facilities for this work are excellent, and the research group has recently won more than $14M for the continuing improvement of its infrastructure. Please contact the Departmental Administrator for further details, at the Neural Systems Group, Department of Psychology, Ridley Building, Claremont Place, Newcastle upon Tyne, NE1 7RU. Tel: +44-191-222-7525, Fax: +44-191-222-5622. E-mail: m.p.young at ncl.ac.uk - From Nello.Cristianini at bristol.ac.uk Fri Dec 17 13:02:18 1999 From: Nello.Cristianini at bristol.ac.uk (N Cristianini) Date: Fri, 17 Dec 1999 18:02:18 +0000 (GMT) Subject: CFP: Support Vector - Special Issue Message-ID: Please post the following Call for Papers thanks Nello -------------------------------------------------------- Call for papers: ================ Special Issue of Machine Learning Journal on SUPPORT VECTOR MACHINES AND KERNEL METHODS Guest editors: Nello Cristianini (Bristol University, UK - contact person) Colin Campbell (Bristol University, UK) Chris Burges (Lucent Technologies, USA) Deadline: March 15th, 2000 More information on the webpage: http://lara.enm.bris.ac.uk/cig/specissue.htm or contact: Nello Cristianini, Department of Engineering Mathematics, Queen's Building, Bristol University, Bristol BS8 1TR United Kingdom Nello.Cristianini at bristol.ac.uk From lemm at lorentz.uni-muenster.de Mon Dec 20 13:29:13 1999 From: lemm at lorentz.uni-muenster.de (Joerg_Lemm) Date: Mon, 20 Dec 1999 19:29:13 +0100 (CET) Subject: Paper on Mixtures of Gaussian Process Priors Message-ID: Dear colleagues, The following paper (10 pages) is now available at http://xxx.lanl.gov/ps/physics/9911077 --------------------------------------------------------------------------- "Mixtures of Gaussian process priors" by Joerg Lemm, Muenster University Abstract: Nonparametric Bayesian approaches based on Gaussian processes have recently become popular in the empirical learning community. They encompass many classical methods of statistics, like Radial Basis Functions or various splines, and are technically convenient because Gaussian integrals can be calculated analytically. Restricting to Gaussian processes, however, forbids for example the implemention of genuine nonconcave priors. Mixtures of Gaussian process priors, on the other hand, allow the flexible implementation of complex and situation specific, also nonconcave "a priori" information. This is essential for tasks with, compared to their complexity, a small number of available training data. The paper concentrates on the formalism for Gaussian regression problems where prior mixture models provide a generalisation of classical quadratic, typically smoothness related, regularisation approaches being more flexible without having a much larger computational complexity. --------------------------------------------------------------------------- The paper and related material can also be found at my home page: http://pauli.uni-muenster.de/~lemm/ ======================================================================== Joerg Lemm Universitaet Muenster Email: lemm at uni-muenster.de Institut fuer Theoretische Physik I Phone: +49(251)83-34922 Wilhelm-Klemm-Str.9 Fax: +49(251)83-36328 D-48149 Muenster, Germany http://pauli.uni-muenster.de/~lemm ======================================================================== From harnad at coglit.ecs.soton.ac.uk Mon Dec 20 17:38:49 1999 From: harnad at coglit.ecs.soton.ac.uk (Stevan Harnad) Date: Mon, 20 Dec 1999 22:38:49 +0000 (GMT) Subject: AUTONOMOUS BRAIN/Milner: PSYC Call for Book Reviewers (673 lines) Message-ID: PSYCOLOQUY CALL FOR BOOK REVIEWERS Below is the Abstract of "The Autonomous Brain" by Peter M Milner This book has been selected for multiple review in Psycoloquy. If you wish to submit a formal book review please write to psyc at pucc.princeton.edu indicating what expertise you would bring to bear on reviewing the book if you were selected to review it. (If you have never reviewed for PSYCOLOQUY or Behavioral & Brain Sciences before, it would be helpful if you could also append a copy of your CV to your inquiry.) If you are selected as one of the reviewers and do not have a copy of the book, you will be sent a copy of the book directly by the publisher (please let us know if you have a copy already). Reviews may also be submitted without invitation, but all reviews will be refereed. The author will reply to all accepted reviews. FULL PSYCOLOQUY BOOK REVIEW INSTRUCTIONS AT: http://www.princeton.edu/~harnad/psyc.html http://www.cogsci.soton.ac.uk/psycoloquy/ Psycoloquy reviews are of the book not the Precis. Length should be about 200 lines [c. 1800 words], with a short abstract (about 50 words), an indexable title, and reviewer's full name and institutional address, email and Home Page URL. All references that are electronically accessible should also have URLs. AUTHOR'S RATIONALE FOR SOLICITING MULTIPLE REVIEW: "I hope this book will attract the attention of psychologists, philosophers and others interested in the mind, to some advances in neuroscience that are relevant to cognition. In particular, I think the idea that the neural processes involved in motivation and intention play an essential role in the learning of percepts should be more widely recognised. I acknowledge that no single person can hope to be well informed about the entire fields of neuroscience and cognition and it is my hope that multiple reviews of the work by experts may expose the inevitable inaccuracies and gaps, providing a partial substitute for multiple authorship. Having a number of reviews in one readily accessible place would provide a valuable resource for the reader." ----------------------------------------------------------------------- psycoloquy.99.10.071.autonomous-brain.1.milner Mon Dec 20 1999 ISSN 1055-0143 (50 paragraphs, 20 references, 620 lines) PSYCOLOQUY is sponsored by the American Psychological Association (APA) Copyright 1999 Peter M Milner THE AUTONOMOUS BRAIN [Lawrence Erlbaum Associates, Mahwah, NJ.155 pp. ISBN: 0-8058-3211-4] Precis of Milner on Autonomous-Brain Peter M. Milner McGill University Department of Psychology 1205 Dr. Penfield Ave. Montreal, QC. CANADA H3A 1B1 ps64 at musica.mcgill.ca ABSTRACT: This book presents a theory of behaviour based on the premise that nervous systems have evolved to enable animals to engage in a variety of useful activities. Information about the outside world is essential for most of these activities, but the theory that sensory input must shape its own information processing system is rejected. After the behaviourists banished the immaterial self from psychology they replaced it by stimulus input, and for many years behaviour was attributed entirely to sensory input. Only recently has the notion started to develop that it is the response mechanism of the brain that determines what stimuli are required to perform an action. In the model presented here, an executive system located in the frontal region of the brain employs the extensive reciprocal connections of the sensory cortex to select the input needed to guide the motor system. The consequences of having an autonomous response planner, instead of one subservient to outside stimuli, are far reaching. Neural representations of broad categories that can coexist with multiple distinct subclasses, and the related phenomenon of stimulus equivalence, become easier to understand, for example. It may also enable us to understand why we usually think that we make our own decisions. The book also has suggestions about the way serial order is learned and the role of the frontal regions of the brain in reinforcement, expectancy and response planning. KEYWORDS: association of ideas, attention, behaviour model, intention, motivation, self, serial order Full text of Precis of book available at: http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.071 or ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.071.autonomous-brain.1.milner From eurich at theo.physik.uni-bremen.de Tue Dec 21 08:43:29 1999 From: eurich at theo.physik.uni-bremen.de (Christian Eurich) Date: Tue, 21 Dec 1999 14:43:29 +0100 Subject: Preprints on the encoding accuracy of neural populations Message-ID: <385F8401.87B52DFA@physik.uni-bremen.de> Dear Connectionists, two preprints on Fisher information and the encoding accuracy of neural populations are available from my homepage at http://www-neuro.physik.uni-bremen.de/~eurich/Publications/publications.html 1. Multi-dimensional encoding strategy of spiking neurons C. W. Eurich and S. D. Wilke Neural Computation (in press) Abstract: Neural responses in sensory systems are typically triggered by a multitude of stimulus features. Using information theory, we study the encoding accuracy of a population of stochastically spiking neurons characterized by different tuning widths for the different features. The optimal encoding strategy for representing one feature most accurately consists of (i) narrow tuning in the dimension to be encoded to increase the single-neuron Fisher information, and (ii) broad tuning in all other dimensions to increase the number of active neurons. Extremely narrow tuning without sufficient receptive field overlap will severely worsen the coding. This implies the existence of an optimal tuning width for the feature to be encoded. Empirically, only a subset of all stimulus features will normally be accessible. In this case, relative encoding errors can be calculated which yield a criterion for the function of a neural population based on the measured tuning curves. 2. Neural representation of multi-dimensional stimuli C. W. Eurich, S. D. Wilke and H. Schwegler Proc. NIPS '99 (in press). Abstract: The encoding accuracy of a population of stochastically spiking neurons is studied for different distributions of their tuning widths. The situation of identical radially symmetric receptive fields for all neurons, which is usually considered in the literature, turns out to be disadvantageous from an information-theoretic point of view. Both a variability of tuning widths and a fragmentation of the neural population into specialized subpopulations improve the encoding accuracy. Both preprints are available in postscript and g-zipped postscript format. ================================================== Dr. Christian Eurich Universitaet Bremen Institut fuer Theoretische Neurophysik, FB 1 Postfach 330 440 D-28334 Bremen Germany Phone: +49 (421) 218-4559 Fax: +49 (421) 218-9104 Phone home: +49 (421) 5577208 Fax home: +49 (421) 5577050 e-mail: eurich at physik.uni-bremen.de homepage: http://www-neuro.physik.uni-bremen.de/~eurich "Die Welt ist alles, was der Fall ist." (L. W.) ================================================== From smagt at dlr.de Tue Dec 21 10:18:28 1999 From: smagt at dlr.de (Patrick van der Smagt) Date: Tue, 21 Dec 1999 16:18:28 +0100 (MET) Subject: CFP: Scalable Robotic Applications of Neural Networks Message-ID: <199912211518.QAA18903@ilz.robotic.dlr.de> From smagt at dlr.de Tue Dec 21 10:18:28 1999 From: smagt at dlr.de (Patrick van der Smagt) Date: Tue, 21 Dec 1999 16:18:28 +0100 (MET) Subject: CFP: Scalable Robotic Applications of Neural Networks Message-ID: <199912211518.QAA18903@ilz.robotic.dlr.de> REMINDER: CALL FOR PAPERS Scalable Robotic Applications of Neural Networks http://www.robotic.dlr.de/Smagt/CFP/ Special Issue of "Applied Intelligence" Submission deadline: January 15, 2000 In this special issue, we want to address the question of whether there are real robotic problems that can be better solved using existing neuro-computational principles than by using other standard engineering techniques. Where sensory-motor systems cannot be explicitly modeled, the brain's success leads us to expect that approaches based on adaptive neural control will someday provide a technically sound alternative. However, today's robotic engineers are appropriately skeptical regarding use of neural network principles because of the apparent scarcity of published research that demonstrates scalability of solutions to complex robotic control tasks. For example, like many traditional approaches, many neural network control strategies do not scale well from a two-degree-of-freedom robot arm to a seven degree-of-freedom system. For this special issue, papers are sought that: 1. introduce or review biologically plausible models of sensory-motor control that truly integrate action and perception; 2. newly apply such models to the control of realistic robots, especially manipulators; 3. describe hybrid robot control methodologies that incorporate cerebellar or other neuro-computational models (e.g., vision); 4. provide a case history that clearly defines, or illustrates overcoming of, barriers to successful competition by neural network models for robot control. The scalability of applications of biological models can only be fully assessed in the context of demonstrations that are representative of real-world complexities. Therefore, results on real hardware will generally be preferred. However, results on well-simulated complex problems will be preferred to results with hardware (e.g., 2 DoF robot arms) that can already be optimally controlled with conventional, non-neural techniques. -------------------------------------- Special Issue Editors Patrick van der Smagt Institute of Robotics and System Dynamics DLR (German Aerospace Research Center) Oberpfaffenhofen, Germany smagt at dlr.de http://www.robotic.dlr.de/Smagt/ Daniel Bullock Cognitive and Neural Systems Department Boston University danb at cns.bu.edu http://cns-web.bu.edu/faculty.html#bullock Editor in Chief Moonis Ali Department of Computer Science Southwest Texas State University ma04 at swt.edu http://www.cs.swt.edu -------------------------------------- Submission Deadline for submitting papers: Jan 15, 2000. The journal's instructions to authors can be read here. For this special issue, for review all papers must be submitted by the above date in Postscript or PDF format via email to smagt at dlr.de. Hardcopy submissions will not be accepted. To the extent that time permits, the editors will invite short external commentaries (to be published in the same issue) on the accepted papers. Submitters should inform the editors if they do *not* want their paper treated in an invited commentary. Suggestions of expert reviewers and commentators are welcome. -------------------------------------- About the journal The objective of Applied Intelligence (IJAI) is to provide a medium for exchanging applied research on intelligent systems and technological achievements. IJAI is currently in its eighth year of publication. In order to meet the demand, the publication frequency has been increased from four to six issues per year effective 1998. IJAI is abstracted and/or indexed by twenty indexing publications. See http://www.wkap.nl/journals/apin. Wishing you all a Merry Christmas and a Happy New Year! Patrick -- Dr Patrick van der Smagt phone +49 8153 281152, fax -34 DLR/Institute of Robotics and Mechatronics smagt at dlr.de P.O.Box 1116, 82230 Wessling, Germany http://www.robotic.de/Smagt/ From stork at rsv.ricoh.com Tue Dec 21 17:18:09 1999 From: stork at rsv.ricoh.com (David G. Stork) Date: Tue, 21 Dec 1999 14:18:09 -0800 (PST) Subject: Open Mind handwriting project Message-ID: ======================================================================== [[Distribute freely]] *** *** *** Request for assistance with Open Mind Handwriting Recognition *** *** *** Led by Dr. Lambert Schomaker of the UNIPEN Foundation (www.UNIPEN.org), the Open Mind Handwriting Recognition project seeks to collect data and develop recognizers of a variety of handwriting data, such as digits, isolated letters, and words in a variety of western languages (primarily English). Handwriting Recognition requires large sets of training data (as do other Open Mind projects) and our hypothesis is that such data can be contributed by non-expert netizens over the web. You can read background material at: www.OpenMind.org/OpenMindICDAR.pdf and www.OpenMind.org/OpenMindICDAR.typeset.pdf In a character-based approach, images of unknown recorded isolated character patterns are presented on a netizen's browser. The netizen then submits to the Open Mind host his or her classification of the character by clicking the corresponding button on a simple interface. For the case of words (having a known ASCII string representation but an unknown location of the letters in the image), the interface is a bit more complex but the general idea remains the same: humans are quite good at pointing out the individual characters in a word. While a small-scale web-based labeling system has been written, there is much software engineering to be done before it is "ready for prime time." Several Linux example programs for labeling are available to serve as an inspiration for developing the browser version for netizens. Issues of browser independence, speed, load optimizing, robustness, algorithms for ensuring data reliability, and the balancing of client/server computation must be addressed before the project is ready for load testing and public deployment. We are seeking software engineers to contribute gratis to this effort. Practical experience with Web-based user interfaces, Java, Javascript, CGI programming and Linux (C language) will be required. A pragmatic view, combined with sensitivity to user issues, would be ideal. A knowledge of pattern recognition research, while desirable, is not needed. If you feel you would like to contribute, please send a short e-mail describing your relevant background to me (not Dr. Schomaker directly). Of course, if you have particular questions, please don't hesitate to contact me. In either case, please write "Open Mind Handwriting" in your header. Many thanks in advance. --David Stork stork at OpenMind.org ======================================================================== | The Open Mind Initiative (www.OpenMind.org) is a collaborative | | framework for developing 'intelligent' software over the internet. | | Open Mind extends traditional open source methodology to allow | | non-expert 'netizens' to contribute data needed for such software. | | | | | | To subscribe to a moderated, low-traffic mailing list on Open Mind, | | send email to majordomo at OpenMind.org with the following message: | | subscribe openmind-general | | | | To be removed from the list, send instead the message: | | unsubscribe openmind-general | ======================================================================== From mtipping at microsoft.com Thu Dec 23 12:41:39 1999 From: mtipping at microsoft.com (Michael Tipping) Date: Thu, 23 Dec 1999 09:41:39 -0800 Subject: Paper announcement: the Relevance Vector Machine Message-ID: <2DCBFADAFCBBD21189D400805F6FA1AB0D334209@RED-MSG-12> Dear Connectionists, A paper on the "relevance vector machine", a Bayesian implementation of sparse kernel regression and classification models (which is to appear in the forthcoming proceedings of NIPS*12), is now available online: ftp://ftp.research.microsoft.com/users/mtipping/rvm_nips.ps.gz Title and abstract are below: -------------------------------------------------------------------------- The Relevance Vector Machine Michael E. Tipping The support vector machine (SVM) is a state-of-the-art technique for regression and classification, combining excellent generalisation properties with a sparse kernel representation. However, it does suffer from a number of disadvantages, notably the absence of probabilistic outputs, the requirement to estimate a trade-off parameter and the need to utilise 'Mercer' kernel functions. In this paper we introduce the Relevance Vector Machine (RVM), a Bayesian treatment of a generalised linear model of identical functional form to the SVM. The RVM suffers from none of the above disadvantages, and examples demonstrate that for comparable generalisation performance, the RVM requires dramatically fewer kernel functions. From steve at cns.bu.edu Fri Dec 24 11:09:52 1999 From: steve at cns.bu.edu (Stephen Grossberg) Date: Fri, 24 Dec 1999 11:09:52 -0500 Subject: neural dynamics of speech perception and word recognition Message-ID: The following article is available at http://www.cns.bu.edu/Profiles/Grossberg in PDF and G'zipped Postscript versions. Grossberg, S. and Myers, C.W. (2000). The resonant dynamics of speech perception: Interword integration and duration-dependent backward effects. Psychological Review, in press. ABSTRACT: How do listeners integrate temporally distributed phonemic information into coherent representations of syllables and words? During fluent speech perception, variations in the durations of speech sounds and silent pauses can produce different perceived groupings. For example, increasing the silence interval between the words "gray chip" may result in the percept "great chip", whereas increasing the duration of fricative noise in "chip" may alter the percept to "great ship" (Repp et al., 1978). The ARTWORD neural model quantitatively simulates such context-sensitive speech data. In ARTWORD, sequential activation and storage of phonemic items in working memory provides bottom-up input to unitized representations, or list chunks, that group together sequences of items of variable length. The list chunks compete with each other as they dynamically integrate this bottom-up information. The winning groupings feed back to provide top-down support to their phonemic items. Feedback stablishes a resonance which temporarily boosts the activation levels of selected items and chunks, thereby creating an emergent conscious percept. Because the resonance evolves more slowly than working memory activation, it can be influenced by information presented after relatively long intervening silence intervals. The same phonemic input can hereby yield different groupings depending on its arrival time. Processes of resonant transfer and competitive teaming help determine which groupings win the competition. Habituating levels of neurotransmitter along the pathways that sustain the resonant feedback lead to a resonant collapse that permits the formation of subsequent resonances. From fayyad at MICROSOFT.com Sat Dec 25 02:55:30 1999 From: fayyad at MICROSOFT.com (Usama Fayyad) Date: Fri, 24 Dec 1999 23:55:30 -0800 Subject: SIGKDD Explorations: vol. 1:2, announcements, news, and advertise ments Message-ID: The first issue of SIGKDD Explorations, the official newsletter of the ACM's new Special Interest Group (SIG) on Knowledge Discovery and Data Mining was published in June 1999 and was sent to all SIGKDD members and ACM network of libraries. It will also be included in SIGMOD's DiSC anthology CD-ROM. If you are a member and did not receive your hardcopy, please e-mail: help at acm.org to fix any problems with your membership. The first vlume is available to all on the web at: http://research.microsoft.com/datamine/SIGKDD. If you have any short news items or announcements approriate for inclusion in the second issue, please submit them to fayyad at acm.org or sunita at cs.berkeley.edu. For more information and examples from issue 1, please see: http://research.microsoft.com/datamine/SIGKDD. SIGKDD Explorations is a twice-annual newsletter dedicated to serve the SIGKDD membership and community. Anoouncements should not exceed 4 lines in length. Please provide URL if there is associated web information. *NEW*: We also have added a for-pay advertisement section to allow vendors, companies, consultants, and others to reach the rapidly growing SIGKDD community. Advertising rates start at $250 for quarter page, $500 per half page, and $1000 for a full page. All submissions must arrive by Jan. 20, 1999 for inclusion in the next issue. thanks, Usama Fayyad, Editor-in-Chief (fayyad at acm.org) Sunita Sarawagi, Associate Editor (sunita at cs.berkeley.edu) http://research.microsoft.com/datamine/SIGKDD From doya at ctr.atr.co.jp Sat Dec 25 06:13:21 1999 From: doya at ctr.atr.co.jp (Kenji Doya) Date: Sat, 25 Dec 1999 20:13:21 +0900 Subject: Research positions: computational neuroscience and robotics Message-ID: <3864A5BE.F93732E1@ctr.atr.co.jp> Post-doctoral Research Positions METALEARNING, NEUROMODULATION AND EMOTION http://meta.rad.atr.co.jp/ CREST, Japan Science and Technology Corporation The goal of the research project is to understand the mechanisms of higher-level regulation of learning functions of the brain. We propose computational models of the roles of the neuromodulators (e.g., dopamine, serotonin, noradrenaline, acetylecholine), test them by neurobiological and clinical experiments, and explore the functions of emotion-like systems in adaptive communication agents. Post-doctoral research positions are available immediately for the studies of 1) modeling the functions of neuromodulators in close collaboration with the experimental laboratories below; 2) development of self regulated learning algorithms for adaptive agents; and 3) designing and building a colony of robots with rich sensory and motor functions. Candidates should have a strong background in theoretical and/or experimental sciences, including neurobiology, psychology, computer science, robotics, physics and mathematics. Competitive salaries and benefits are provided by JST (Japan Science and Technology Corporation). The laboratory is located at ATR International in Kyoto, where people from diverse origins and backgrounds get together for basic research on human cognition and telecommunication. The research project is supported by JST's "Creating the Brain" program, supervised by Dr. Shun-ichi Amari. The principal project members and their research topics are the following: Takashi Matsumoto (Waseda University): Development of on-line Bayesian learning theory and algorithms. Shin Ishii (Nara Institute of Science and Technology): Development of metalearning algorithms and architectures for adaptive agents. Yuko Sekino (Gunma University): In vivo and in vitro experiments on regulation of hippocampal memory by neuromodulators in rats and mice. Toshiyuki Sawaguchi (Hokkaido University): Experiments on the roles of neuromodulators in the prefrontal cortex in monkeys. Shigeto Yamawaki (Hiroshima University): Behavioral and pharmacological studies on the roles of serotonin in human cognition using fMRI, MEG and TMS. Kenji Doya (ATR International): Development of a computational theory of neuromodulator functions; Real and simulated experiments of learning and communication in multiple agents. Applicants should send a CV, a statement of research interests, and recent reprints and/or preprints by January 31st, 2000 to Kenji Doya ATR International 2-2 Hikaridai, Seika, Soraku Kyoto 619-0288, Japan For more information, please contact info at meta.rad.atr.co.jp or visit our web page at http://meta.rad.atr.co.jp/ -- Kenji Doya ATR International 2-2 Hikaridai, Seika, Soraku, Kyoto 619-0288, Japan Phone:+81-774-95-1251, Fax:+81-774-95-1259 E-mail:doya at ctr.atr.co.jp, http://meta.rad.atr.co.jp/doya/ From achim at cse.unsw.edu.au Wed Dec 1 03:57:50 1999 From: achim at cse.unsw.edu.au (Achim Hoffmann) Date: Wed, 1 Dec 99 19:57:50 +1100 Subject: Book Announcement: Paradigms of Artificial Intelligence, Springer-Verlag Message-ID: <991201085750.15517@cse.unsw.edu.au> My apologies, if you receive this announcement more than once. Paradigms of Artificial Intelligence A Methodological and Computational Analysis by Achim Hoffmann, Springer-Verlag 1998 http://www.cse.unsw.edu.au/~achim/Book/ From aida at kouku-k.ac.jp Thu Dec 2 03:20:13 1999 From: aida at kouku-k.ac.jp (Toshiaki Aida) Date: Thu, 02 Dec 1999 17:20:13 +0900 Subject: Paper Available Message-ID: <38462BBD.C6E43C05@kouku-k.ac.jp> Dear colleagues, The following paper is now available at http://xxx.lanl.gov/abs/cond-mat/9911474 , in which we discussed the control of bin size for optimal on-line learning of probability distributions. TITLE: Field Theoretical Analysis of On-line Learning of Probability Distributions, Toshiaki Aida, Physical Review Letters 83 (1999) 3554-3557. ABSTRACT: On-line learning of probability distributions is analyzed from the field theoretical point of view. We can obtain an optimal on-line learning algorithm, since renormalization group enables us to control the number of degrees of freedom of a system according to the number of examples. We do not learn parameters of a model, but probability distributions themselves. Therefore, the algorithm requires no a priori knowledge of a model. Best regards, Toshiaki Aida Department of Physics, Tokyo Institute of Technology, Department of Aeronautics, Tokyo Metropolitan College of Aeronautical Engineering, E-mail: aida at stat.phys.titech.ac.jp From maarten at swi.psy.uva.nl Thu Dec 2 07:39:19 1999 From: maarten at swi.psy.uva.nl (Maarten van Someren) Date: Thu, 2 Dec 1999 13:39:19 +0100 (MET) Subject: MLNET - for your mailinglist Message-ID: MLNET - a European network of excellence in Machine Learning, Case-Based Reasoning and Knowledge Acquisition MLNET is a "network of excellence" in Machine Learning that is funded by the European Community to coordinate research, development and application of Machine Learning, Case-Based Reasoning and Knowledge Acquisition in the European Community. Activities of the network are: * development and maintenance of an electronic information service that gives access to information on: research groups, projects, datasets, software, educational materials and links to other information services (see http://www.mlnet.org/). This is the successor to the GMD information service developed earlier by MLNET. * development and maintenance of a scientific and technological outlook ("technological roadmap") for these areas that acts as a guideline for the planning and coordination of research and development. (Members of MLNET can take part in developing this.) * development and maintenance of educational and "advertising" materials for the areas of the network. * organisation and support of events (e.g. European Conference on Machine Learning, European Workshop on Case-Based Reasoning, European Knowledge Acquisition Workshop, Multi-strategy learning workshop). Members of MLNET can propose to organise such events and apply for support. * funding of "ambassadors" who give presentations on Machine Learning, Case-Based Reasoning or Knowledge Acquisition OUTSIDE the areas. (proposals for "ambassador visits" by a member of MLNET can be submitted to the coordinator of MLNET, see below). Institutes and companies in the European community and associated states can apply for membership. At the moment the information service is available for everyone. Only members of MLNET can receive financial support for organising or visiting MLNET events. MLNET collaborates with three other European networks (ERUDIT on uncertainty modelling, EvoNet on evolutionary computing and NEuroNet on neural network computing) on the theme Computational Intelligence and Learning (see http://www.dcs.napier/coil/). More information can be found at the MLNET Information Service. - maarten van someren (coordinator of MLNET, maarten at swi.psy.uva.nl) From rsun at cecs.missouri.edu Sat Dec 4 19:12:56 1999 From: rsun at cecs.missouri.edu (Ron Sun) Date: Sat, 4 Dec 1999 18:12:56 -0600 Subject: Papers on cognitive modeling using reinforcement learning Message-ID: <199912050012.SAA05951@pc113.cecs.missouri.edu> Announcing four papers on cognitive modeling and cognitive architectures based on hybrid reinforcement learning methods --- the CLARION model: A paper on cognitive modeling using CLARION: -------------------------------------------------- From piuri at elet.polimi.it Sat Dec 4 11:04:14 1999 From: piuri at elet.polimi.it (Vincenzo Piuri) Date: Sat, 04 Dec 1999 17:04:14 +0100 Subject: IJCNN'2000: CFP & Submission of extended versions to IEEE Transactions on Neural Networks!!!! Message-ID: <3.0.5.32.19991204170414.0151ac60@elet.polimi.it> ======================================================================== IEEE-INNS-ENNS INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS Grand Hotel di Como, Como, Italy - 24-27 July 2000 PAPER SUBMISSION DEADLINE is coming soon!!!! Please, have a look to the call for papers in the conference web site. Do not miss the opportunity to submit a paper and participate actively to the conference! CALL FOR SPECIAL SESSIONS AND TUTORIALS have been also published. You can find them on the conference web site. NEW!!!! NEW!!!! NEW!!!!! SUBMISSION OF AN EXTENDED VERSION OF THE PAPER TO IEEE-TNN: The authors of selected IJCNN'2000 papers will be invited to submit an extended version to the IEEE Transactions on Neural Networks by 30 SEPTEMBER 2000. These submitted papers will undergo the usual Transactions review process. However, the Transactions Editors will receive the IJCNN'2000 reviewing reports for each selected IJCNN'2000 paper, and thus the reviewing process is likely to be simpler and faster, although accurate as usual. For each paper, the Associate Editor in charge may in fact ask the IJCNN'2000 reviewers (and/or PC members) to provide an in-depth review of the extended version for the IEEE-TNN. Since the reviewers and the PC member had already evaluated a preliminary version of the paper, their work is likely to be simplified. official conference web site: http://www.ims.unico.it/2000ijcnn.html conference mirror web site: http://www.lans.ece.utexas.edu/2000ijcnn.html ====================================================================== Vincenzo Piuri Department of Electronics and Information, Politecnico di Milano piazza L. da Vinci 32, 20133 Milano, Italy phone +39-02-2399-3606 secretary +39-02-2399-3623 fax +39-02-2399-3411 email piuri at elet.polimi.it From nnsp00 at neuro.kuleuven.ac.be Fri Dec 3 11:27:07 1999 From: nnsp00 at neuro.kuleuven.ac.be (NNSP2000, Sydney) Date: Fri, 03 Dec 1999 17:27:07 +0100 Subject: IEEE workshop on Neural Networks for Signal Processing (NNSP), Sydney, Australia, December 2000. Message-ID: <3847EF5B.CF3908CB@neuro.kuleuven.ac.be> Please find attached advanced information about the IEEE workshop on Neural Networks for Signal Processing (NNSP) which is to be held in Sydney, Australia, in December 2000. In case you would like to be removed from our mailing list: reply to this mail with as subject "remove". Marc M. Van Hulle Katholieke Universiteit Leuven Belgium *********************************************** **** CALL FOR PAPERS **** **** submission deadline: March 31, 2000 **** *********************************************** December 11-13, 2000, Sydney, Australia NNSP'2000 homepage: http://eivind.imm.dtu.dk/nnsp2000 Thanks to the sponsorship of IEEE Signal Processing Society and the IEEE Neural Networks Council the tenth of a series of IEEE workshops on Neural Networks for Signal Processing will be held at the University of Sydney Campus, Sydney, Australia. The workshop will feature keynote lectures, technical presentations, and panel discussions. Papers are solicited for, but not limited to, the following areas: Algorithm and Architectures: Artificial neural networks (ANN), adaptive signal processing, Bayesian modeling, MCMC, generalization, design algorithms, optimization, parameter estimation, nonlinear signal processing, Markov models, fuzzy systems (FS), evolutionary computation (EC), synergistic models of ANN/FS/EC, and wavelets. Applications: Speech processing, image processing, sonar and radar, data fusion, intelligent multimedia and web processing, OCR, robotics, adaptive filtering, blind source separation, communications, sensors, system identification, and other general signal processing and pattern recognition applications. Implementations: Parallel and distributed implementation, hardware design, and other general implementation technologies. PAPER SUBMISSION PROCEDURE Prospective authors are invited to submit a full paper using the electronic submission procedure described at the workshop homepage: http://eivind.imm.dtu.dk/nnsp2000 Accepted papers will be published in a hard-bound volume by IEEE and distributed at the workshop. SCHEDULE Submission of full paper: March 31, 2000 Notification of acceptance: May 31, 2000 Submission of photo-ready accepted paper: July 15, 2000 Advanced registration, before: September 15, 2000 ORGANIZATION Honorary Chair Bernard WIDROW Stanford University General Chairs Ling GUAN University of Sydney email: ling at ee.usyd.edu.au Kuldip PALIWA Griffith University email: kkp at shiva2.me.gu.edu.au Program Chairs Tulay Adali University of Maryland, Baltimore County email: adali at umbc.edu Jan LARSEN Technical University of Denmark email: jl at imm.dtu.dk Finance Chair Raymond Hau-San WONG University of Sydney email: hswong at ee.usyd.edu.au Proceedings Chairs Elizabeth J. WILSON Raytheon Co. email: bwilson at ed.ray.com Scott C. DOUGLAS Southern Methodist University email: douglas at seas.smu.edu Publicity Chair Marc Van HULLE Katholieke Universiteit, Leuven email: marc at neuro.kuleuven.ac.be Registration and Local Arrangements Stuart PERRY Defence Science and Technology Organisation email: Stuart.Perry at dsto.defence.gov.au Europe Liaison Jean-Francois CARDOSO ENST email: cardoso at sig.enst.fr America Liaison Amir ASSADI University of Wisconsin at Madison email: ahassadi at facstaff.wisc.edu Asia Liaison Andrew BACK RIKEN email: andrew.back at usa.net Program Committee Amir Assadi Yianni Attikiouzel John Asenstorfer Andrew Back Geoff Barton Herve Bourlard Andy Chalmers Zheru Chi Andrzej Cichocki Tharam Dillon Tom Downs Hsin Chia Fu Suresh Hangenahally Marwan Jabri Haosong Kong Shigeru Katagiri Anthony Kuh Yi Liu Fa-Long Luo David Miller Christophe Molina M Mohammadian Erkki Oja Soo-Chang Pei Jose Principe Ponnuthurai Suganthan Ah Chung Tsoi Marc Van Hulle A.N. Venetsanopoulos Yue Wang From m.niranjan at dcs.shef.ac.uk Fri Dec 3 03:34:08 1999 From: m.niranjan at dcs.shef.ac.uk (Mahesan Niranjan) Date: Fri, 3 Dec 1999 08:34:08 +0000 (GMT) Subject: Faculty Openings at Sheffield Message-ID: <199912030834.IAA00602@bayes.dcs.shef.ac.uk> Dear Connectionists, A number of faculty positions are being currently advertised in the Department of Computer Science in Sheffield. For one of these, we are looking for someone in the area of Machine Learning - this post was created with my appointment here, and the target candidate will be someone with research interest in Probabilistic Modelling with a signal processing bias. UK academic jobs pay low and are increasingly stressful due to falling Maths standards of students coming in. But this is still a nice job! Please visit http://www.dcs.shef.ac.uk if you might be interested, or pass this to anyone you think might be interested. Closing date 20 December 1999. Many thanks niranjan ____________________________________________________________________ Mahesan Niranjan Phone: 44 114 222 1805 Professor of Computer Science FaX: 44 114 222 1810 The University of Sheffield Email: M.Niranjan at dcs.shef.ac.uk http://www.dcs.shef.ac.uk/~niranjan ____________________________________________________________________ From mdorigo at ulb.ac.be Mon Dec 6 12:17:48 1999 From: mdorigo at ulb.ac.be (Marco DORIGO) Date: Mon, 6 Dec 1999 18:17:48 +0100 Subject: ANTS'2000: Call for papers Message-ID: ANTS'2000 - From Ant Colonies to Artificial Ants: Second International Workshop on Ant Algorithms Brussels, Belgium, September 8-9, 2000 CALL FOR EXTENDED ABSTRACTS (up-to-date information on the workshop is maintained at http://iridia.ulb.ac.be/~ants/ants2000/) SCOPE OF THE WORKSHOP The behavior of social insects in general, and of ant colonies in particular, has since long time fascinated researchers in ethology and animal behavior, who have proposed many models to explain their capabilities. Recently, ant algorithms have been proposed as a novel computational model that replaces the traditional emphasis on control, preprogramming, and centralization with designs featuring autonomy, emergence, and distributed functioning. These designs are proving flexible and robust, able to adapt quickly to changing environments and to continue functioning even when individual elements fail. A particularly successful research direction in ant algorithms, known as "ant colony optimization", is dedicated to their application to discrete optimization problems. Ant colony optimization has been applied successfully to a large number of difficult combinatorial problems like the quadratic assignment and the traveling salesman problems, to routing in telecommunications networks, to scheduling problems, and so on. ANTS'2000 is the second edition of the only event entirely devoted to ant algorithms and to ant colony optimization. Also of great interest to the workshop are models of ant colonies behavior which could stimulate new algorithmic approaches. The aims are to give researchers in both real ants behavior and in ant colony optimization an opportunity to meet, to present their latest research, and to discuss current developments and applications. The two-day event will be held in Brussels, Belgium, from September 8 to September 9, 2000. In the evening of September 7 there will be a tutorial on ant algorithms. RELEVANT RESEARCH AREAS Submissions are invited that describe: (1) Models of aspects of real ant colonies behavior that can stimulate new algorithmic approaches, (2) Empirical and theoretical research in ant algorithms and ant colony optimization. (3) Application of ant algorithms and ant colony optimization methods to real-world problems. SUBMISSION OF PAPERS Papers can be submitted either as extended abstracts or as full papers. The length should be of approximately 4 or 10 pages respectively. Formatting and submission instructions are provided at http://iridia.ulb.ac.be/~ants/ants2000/ants2000-submission.html. Accepted papers will be published in a booklet that is to be distributed to workshop participants. We are now investigating the possibility of publishing the best papers as a journal special issue. In this case, a further call for full papers, that may be longer than the 10 pages workshop format, will be released. All accepted papers will be presented at the workshop as oral presentations or as a poster. REGISTRATION AND FURTHER INFORMATION Up-to-date information about the workshop will be made available at the ANTS'2000 web site (http://iridia.ulb.ac.be/~ants/ants2000/). For information about local arrangements, registration forms, etc., please refer to the above mentioned web site, or contact the local organizer at the address below. A registration fee of BF 7,000 will cover local organization expenses, coffee breaks, and a social dinner on Friday 8 evening. PhD students reduced registration fee: BF 6,000. A proof of inscription is required in that case. IMPORTANT DATES Submission deadline April 15, 2000 Notification of acceptance May 30, 2000 Camera ready copy June 30, 2000 Conference September 8-9, 2000 ANTS'98 CONFERENCE COMMITTEE PROGRAM CHAIR AND LOCAL ARRANGEMENTS Marco DORIGO, IRIDIA, Universite' Libre de Bruxelles, Belgium PROGRAM COMMITTEE Emile AARTS, Philips Research Labs, Eindhoven, The Netherlands Hugues BERSINI, IRIDIA, Universite' Libre de Bruxelles, Belgium Eric BONABEAU, EuroBios, Paris, France Freddy BRUCKSTEIN, Technion, Haifa, Israel Alberto COLORNI, Politecnico di Milano, Italy David CORNE, University of Reading, Great Britain Jean-Louis DENEUBOURG, Universite' Libre de Bruxelles, Belgium Gianni DI CARO, ATR, Kyoto, Japan Marco DORIGO, IRIDIA, Universite' Libre de Bruxelles, Belgium David FOGEL, Natural Selection, La Jolla, CA, USA Luca M. GAMBARDELLA, IDSIA, Lugano, Switzerland Michel GENDREAU, Universite' de Montreal, Canada Deborah GORDON, Stanford University, CA, USA Walter GUTJAHR, University of Vienna, Austria Owen HOLLAND, University of West England, Bristol, UK Vittorio MANIEZZO, Universita' di Bologna, Italy Zbigniew MICHALEWICZ, University of North Carolina, USA Martin MIDDENDORF, Universitaet Karlsruhe, Germany Thomas STUETZLE, Technische Universitaet Darmstadt, Germany Eric TAILLARD, IDSIA, Lugano, Switzerland Guy THERAULAZ, Universite' Paul Sabatier, Toulouse, France Gilles VENTURINI, Universite' de Tours, France SUBMISSIONS ADDRESS Submitted papers should be emailed to: mmi at aifb.uni-karlsruhe.de and to ants at iridia.ulb.ac.be PROGRAM CHAIR ADDRESS Marco Dorigo, Ph.D. Chercheur Qualifie' du FNRS Tel +32-2-6503169 IRIDIA CP 194/6 Fax +32-2-6502715 Universite' Libre de Bruxelles Secretary +32-2-6502729 Avenue Franklin Roosevelt 50 http://iridia.ulb.ac.be/dorigo/dorigo.html 1050 Bruxelles http://iridia.ulb.ac.be/~ants/ants2000/ Belgium http://iridia.ulb.ac.be/dorigo/ACO/ACO.html CONFERENCE LOCATION Avenue A. Buyl 87 Building C - 4th floor (IRIDIA, Dorigo's lab, is at the 5th floor of the same building). There will be arrows and indications to the workshop room. LIMITED NUMBER OF PLACES The number of participants will be limited. If you intend to participate please fill in and send the intention form available at the workshop web page, or drop a line to ants at iridia.ulb.ac.be. RELATED CONFERENCES Note that just after ANTS'2000 there will be in Paris (85 minutes by train from Brussels) two related conferences: SAB2000, September 11-15, and PPSN-2000, September 16-20. From K.Althoefer at kcl.ac.uk Mon Dec 6 13:45:24 1999 From: K.Althoefer at kcl.ac.uk (Althoefer, Kaspar) Date: Mon, 06 Dec 1999 18:45:24 +0000 Subject: Research Studentship in the Dept. of Mechanical Engineering, King's College London References: <36FB78FF.BEE3684A@kcl.ac.uk> Message-ID: <384C0444.73B9F919@kcl.ac.uk> Dear Colleagues, I have a Research Studentship available here in London, UK, to work on motion planning and control algorithms for robots involving research on sensors and neural networks. Applicants should hold a first class honours degree in mechanical engineering, electronic engineering or a related subject. The studentship is only open for candidates from the European Union (EU). Further details can be obtained by writing or e-mailing to me. If you know of anyone interested please feel free to pass this information on. Best regards, Kaspar Althoefer. -- |_/ I N G'S Dr Kaspar ALTHOEFER | \ COLLEGE Ph.D., Dipl.-Ing., AMIEE L O N D O N Department of Mechanical Engineering Founded1829 King's College, Strand, London WC2R 2LS, UK TEL: +44 (0)20 7848 2431, FAX: +44 (0)20 7848 2932 e-mail: K.Althoefer at kcl.ac.uk http://www.eee.kcl.ac.uk/~kaspar From mozer at cs.colorado.edu Tue Dec 7 13:31:21 1999 From: mozer at cs.colorado.edu (Mike Mozer) Date: Tue, 07 Dec 99 11:31:21 -0700 Subject: Positions in machine learning, statistics, and data mining Message-ID: <199912071831.LAA11256@neuron.cs.colorado.edu> Athene Software, Inc. Positions in Machine Learning, Statistics, and Data Mining Athene Software, based in Boulder, Colorado, has immediate openings for professionals in machine learning, statistics, and data mining. We are seeking qualified candidates to develop and enhance models of subscriber behavior for telecommunications companies. Responsibilities include: statistical investigation of large data sets, building predictive and decision-making models using the latest advances in machine learning techniques, developing and tuning data representations, and presentation of results to internal and external customers. Candidates must hold a Ph.D. in Computer Science, Statistics, Electrical Engineering, or related field. The ideal candidate will have experience in pattern recognition or mathematical modeling on real world problems, familiarity with experimental design and data analysis, and some background in relational database systems. Strong communication skills are extremely important. Athene has a long-term committment to cultivating a dynamic, stimulating environment for its Ph.D. research staff. The group is slated to double over the next few years. Athene encourages publication of research results and active participation in the research community. And Athene has established a research advisory board consisting of leaders in machine learning, including Dr. Satinder Singh Baveja (AT&T Labs - Research), Prof. Geoffrey Hinton (University College London), Prof. John Moody (OGI), Prof. Andrew Moore (CMU), and Prof. Michael Mozer (Boulder). Send applications to: Dr. Robert Dodier Athene Software, Inc. 2060 Broadway, Suite 300 Boulder, CO 80302 email: robertd at athenesoft.com company URL: www.athenesoft.com From AJain at cc.ucsf.edu Tue Dec 7 13:55:09 1999 From: AJain at cc.ucsf.edu (Jain, Ajay) Date: Tue, 7 Dec 1999 10:55:09 -0800 Subject: Research Positions at UCSF: Computational biology, bioinformatics, databases Message-ID: <112CB72C222BD2118FFF00A0C9D18C6460AEF6@cc.ucsf.edu> ABOUT THE POSITIONS: We are seeking two individuals for post-doctoral positions in computational research applied to the study of cancer. Candidates for these positions must have a PhD in computer science or closely related field and they ideally will have experience applying sophisticated computation to real-world problems (e.g. computational biology, drug discovery, object recognition, robotics, etc.). Experience in machine-learning, neural networks, computational geometry, or physical modeling is beneficial, as is formal training in chemistry, biology, or physics. Experience in applying machine learning techniques to problems involving noisy data is particularly relevant. Candidates will be expected to learn aspects of disciplines complementary to their own and collaborate extensively with clinical and biological researchers. We are also seeking an individual for a challenging database development position. This position is responsible for design and implementation of database and data processing systems to support clinical and bench science efforts in a web-based environment. Designing databases to track human tissue samples, databases for long-term patient follow-up, and deploying nationally accessible biological data repositories all fall under the responsibilities of this position. The systems must be optimized for real-time content retrieval via user-friendly Web interfaces. This position is for a senior developer able to take on a leadership role, supervising and mentoring other developers. Candidates for all of these positions must have excellent communication skills, excellent work history and references, and the ability to work both independently and as part of a multidisciplinary team of scientists, including colleagues from biology, medicine, chemistry, genetics, and other fields. We offer competitive salaries and benefits. UC is an affirmative action and equal opportunity employer. ABOUT THE PLACE (see http://cc.ucsf.edu): Housed within one of the nation's top biomedical research universities, the UCSF Cancer Center brings together hundreds of researchers and clinicians dedicated to four fundamental pursuits: basic science research into the causes and events of cancer's progression; clinical research to translate new knowledge into viable treatments; sensitive, state-of-the-art patient care; and programs in epidemiology, prevention, and cancer control. Basic scientists and clinical health care providers work together toward the translation of promising ideas into the creation of new and better treatments for cancer patients everywhere. ABOUT THE RESEARCH (see http://cc.ucsf.edu/jain): Cancer is a disease marked by accumulating genetic abnormalities, ultimately resulting in malignancy. These abnormalities are evident in the genomic DNA, messenger RNA, and in the expressed proteins of cancer cells. Modern techniques for in vitro genomic and proteomic analysis are generating data on a very large scale (up to 30,000 data points per experiment). DNA microarrays, comparative genomic hybridization, flourescent in situ hybridization, 2D gel electrophoresis, and other techniques produce quantitative data that contain information relevant to the biological state of cells. These data offer an enormous opportunity for systematic computational analyses to make an impact on both therapeutic target discovery as well as in clinical decision-making. There is a pressing need for computational methods to move beyond data visualization and toward generating conclusions that are quantitatively supportable. Coupled with data emerging from the Human Genome Project and modern structure-based drug design techniques, there are significant opportunities to advance cancer research and treatment. Members of the growing Jain Lab are engaged in developing and applying novel computational techniques for: 1) Analyzing biological data gathered from human and animal tumors generated within the Center and with external collaborators. 2) Relating DNA/protein sequences to protein function. 3) Structure-based small-molecule drug design aimed at discovery or refinement of cancer therapeutics. We are seeking energetic individuals to join in our research efforts. ABOUT DR. AJAY N. JAIN: Dr. Jain received his PhD in 1991 from Carnegie Mellon University working with Dr. Dave Touretzky and Dr. Alex Waibel. From 1992 through mid-1999, he worked in a series of start-up biopharmaceutical companies on computational methods for structure-based drug design. Dr. Jain is Director of Informatics at the UCSF Cancer Center and holds faculty appointments in the Cancer Research Institute and Dept. of Laboratory Medicine (see http://cc.ucsf.edu/people/jain_ajay.html). TO APPLY: Please send a full CV and letter describing your qualifications by Fax, US Mail, or email (pdf, plain text, or MS Word). Email is the preferred mode. Contact information follows. ____________________________________________________ Ajay N. Jain, PhD Director of Informatics, UCSF Cancer Center Box 0128, San Francisco, CA 94143-0128 Deliveries: 2340 Sutter St., Room S336, San Francisco Tel: (415) 502-7242 Fax: (415) 476-8218 Email: ajain at cc.ucsf.edu Web: http://cc.ucsf.edu/jain From m.towsey at qut.edu.au Tue Dec 7 21:52:30 1999 From: m.towsey at qut.edu.au (Michael Towsey) Date: Wed, 08 Dec 1999 12:52:30 +1000 Subject: Special Issue of "Applied Intelligence" Message-ID: <384DC7EE.21BD18AE@fit.qut.edu.au> CALL FOR CONTRIBUTIONS The submission of papers is invited for a special issue on "Neural Networks and Machine Learning for Natural Language Processing" in "Applied Intelligence: The International Journal of Artificial Intelligence, Neural Networks, and Complex Problem-Solving Techniques" GUEST EDITORS: Joachim Diederich Claudia Brugman Michael Towsey ISSUE THEME: At the core of several disciplines is the question of natural language processing, and its relationship, mediated by language acquisition, to innate linguistic knowledge. Studies in neurophysiology, as well as recent connectionist natural language processing models, have challenged the "poverty of the stimulus" argument and the existence of "Universal Grammar" as innate grammatical structure. Classical generative linguistic theory assumes a modular, encapsulated and autonomous representation of grammatical knowledge which is independent of other representations. This assumption is consistent with the view that grammatical knowledge is genetically programmed and pre-wired in the brain and implies at the level of evolution a possible sudden emergence of linguistic competence, a capacity which is not shared by other species. Related to the issues of innateness and acquisition is the issue of language universals. As described above, "Universal Grammar" is characterised in terms of innate and modular structures. The relationship between this notion of universal and the empiricist notion of language universal, in terms of properties all languages share, has also been the object of hypothesis and study, and both must be accounted for in any model of the innate aspects of language competence. Moreover, any theory that addresses the neurobiological basis of language acquisition must address the question of how the brain can realise the computation required for language processing. We invite articles which address any of these related issues from the point of view of neural network representation or more general, machine learning for natural language processing. At the core of the research topics outlined above is the question of how much prior knowledge is necessary for natural language learning and how this question can be addressed by theoretical or empirical studies. The theme of this special issue comprises * the investigation of the underlying theoretical foundations, including computational learning theory * the implementation and evaluation of methods for learning natural language with neural networks or other machine learning approaches, and * applications of such learning systems in various domains. TOPICS OF INTEREST The list below gives some examples of intended topics: * spoken language acquisition * connectionist approaches for learning the syntax and semantics of natural languages * statistical approaches for language acquisition * hybrid systems (symbolic machine learning/neural network) for learning natural languages * genetic algorithms for language acquisition * inductive learning for language acquisition, including decision-tree learners and inductive logic programming. * comparative evaluations of different machine learning methods SUBMISSION PROCESS Prospective authors should send an electronic mail message by 10th January 2000 indicating their intent to submit a paper to the guest editor of the special issue, Prof. Joachim Diederich (joachim at fit.qut.edu.au). This message should contain a preliminary abstract and three to five keywords. Six hard copies of the final manuscript should be sent to the guest editor (not to the Applied Intelligence Editorial office) at the following address: Professor Joachim Diederich Machine Learning Research Centre Faculty of Information Technology Queensland University of Technology GPO 2434 Brisbane 4001 QLD Australia To speed up the reviewing process, authors should also send a PostScript version of the paper via email to the guest editor. Prospective authors can find further information about the journal on the home page http://kapis.www.wkap.nl/journalhome.htm/0924-669X SCHEDULE: Expressions of interest: January 10, 2000 Paper submission deadline: March 31, 2000 Review decision by: May 31, 2000 Final manuscript due: July 31, 2000 _________________________________________________ Submitted by Michael Towsey From yoram at stat.Berkeley.EDU Wed Dec 8 13:26:52 1999 From: yoram at stat.Berkeley.EDU (YORAM Gat) Date: Wed, 8 Dec 1999 10:26:52 -0800 (PST) Subject: Paper on generalization bounds for the support vector method Message-ID: The following paper is now available at http://www.stat.Berkeley.EDU/tech-reports/548.ps.Z A bound concerning the generalization ability of a certain class of learning algorithms Yoram Gat Abstract: A classifier is said to have good generalization ability if it performs on test data almost as well as it does on the training data. The main result of this paper provides a sufficient condition for a learning algorithm to have good finite sample generalization ability. This criterion applies in some cases where the set of all possible classifiers has infinite VC dimension. We apply the result to prove the good generalization ability of support vector machines. From terry at salk.edu Wed Dec 8 23:39:36 1999 From: terry at salk.edu (terry@salk.edu) Date: Wed, 8 Dec 1999 20:39:36 -0800 (PST) Subject: COMPUTATIONAL NEUROBIOLOGY GRADUATE PROGRAM Message-ID: <199912090439.UAA01824@hebb.salk.edu> DEADLINE: JANUARY 7, 2000 COMPUTATIONAL NEUROBIOLOGY GRADUATE PROGRAM Department of Biology -- University of California, San Diego http://www.biology.ucsd.edu/compneuro/ The goal of the Computational Neurobiology Graduate Program at UCSD is to train a new generation of researchers who are equally at home measuring large-scale brain activity, analyzing the data with advanced computational techniques, and developing new models for brain development and function. Candidates from a wide range of backgrounds are invited to apply, including Biology, Psychology, Computer Science, Physics and Mathematics. The three major themes in the training program are: 1. Neurobiology of Neural Systems -- Anatomy, physiology and behavior of systems of neurons. Using modern neuroanatomical, neuropharmacological and electrophysiological techniques. Lectures, wet laboratories and computer simulations, as well as research rotations. Major new imaging and recording techniques also will be taught, including two-photon laser scanning microscopy and functional magnetic resonance imaging (fMRI). 2. Algorithms and Realizations for the Analysis of Neuronal Data -- New algorithms and techniques for analyzing data obtained from physiological recording, with an emphasis on recordings from large populations of neurons with imaging and multielectrode recording techniques. New methods for the study of co-ordinated activity, such as multi-taper spectral analysis and Independent Component Analysis (ICA). 3. Neuroinformatics, Dynamics and Control of Systems of Neurons -- Theoretical aspects of single cell function and emergent properties as many neurons interact among themselves and react to sensory inputs. A synthesis of approaches from mathematics and physical sciences as well as biology will be used to explore the collective properties and nonlinear dynamics of neuronal systems, as well as issues of sensory coding and motor control. Requests for application materials should be sent to the Graduate Admissions Office, Department of Biology 0348, 9500 Gilman Drive, UCSD, La Jolla, CA, 92093-0348: [gradprog at biology.ucsd.edu]. The deadline for completed application materials, including letters of reference, is JANUARY 7, 2000. More information about applying to the UCSD Biology Graduate Program: http://www-biology.ucsd.edu/sa/Admissions.html The Biology Department home page is located at: http://www-biology.ucsd.edu/ Other inquiries about the Computational Neurobiology Graduate Program should be directed to: Terrence Sejnowski Institute for Neural Computation 0523 University of California, San Diego La Jolla, CA 92093 tsejnowski at ucsd.edu Participating Faculty include: * Henry Abarbanel (Physics): Nonlinear and oscillatory dynamics; modeling central pattern generators in the lobster stomatogastric ganglion. Director of the Institute for Nonlinear Systems at UCSD. * Thomas Albright (Salk Institute): Motion processing in primate visual cortex; linking single neurons to perception; fMRI in awake, behaving monkeys. Director, Sloan Center for Theoretical Neurobiology. * Darwin Berg (Biology): Regulation synaptic components, assembly and localization, function and long-term stability. Former Chairman of Biology. * Garrison Cottrell (Computer Science and Engineering): Dynamical neural network models and learning algorithms. * Mark Ellisman (Neurosciences, School of Medicine): High resolution electron and light microscopy; anatomical reconstructions. Director, National Center for Microscopy and Imaging Research. * Robert Hecht-Nielsen (Electrical and Computer Engineering): Neural computation and the functional organization of the cerebral cortex. Founder of Hecht-Nielsen Corporation. * Harvey Karten (Neurosciences, School of Medicine): Anatomical, physiological and computational studies of the retina and optic tectum of birds and squirrels. * David Kleinfeld (Physics):Active sensation in rat somatosensation; properties of neuronal assemblies; optical imaging of large-scale activity. Co-director, Analysis of Neural Data Workshop (MBL). * William Kristan (Biology): Neuroethology of leech; functional and developmental studies of the leech nervous system, including computational studies of the bending reflex and locomotion. Director of the Neurosciences Graduate Program. * Herbert Levine (Physics): Nonlinear dynamics and pattern formation in physical and biological systems, including cardiac dynamics and the growth and form of bacterial colonies. * Javier Movellan (Cognitive Science): Sensory fusion and learning algorithms for continuous stochastic systems. * Mu-ming Poo (Biology): Mechanisms for synaptic plasticity; developmental plasticity and learning in nervous systems; development of sensory maps in lower vertebrate visual systems. * Mikhael Rabinovich (Institute for Nonlinear Science): Dynamical systems analysis of the stomatogastric ganglion of the lobster and the antenna lobe of insects. * Terrence Sejnowski (Salk Institute/Biology): Computational neurobiology; physiological studies of neuronal reliability and synaptic mechanisms. Director, Institute for Neural Computation. * Martin Sereno (Cognitive Science): Neural bases of visual cognition and language using anatomical, electrophysiological, computational, and non-invasive brain imaging techniques. * Nicholas Spitzer (Biology): Regulation of ionic channels and neurotransmitters in neurons; effects of electrical activity in developing neurons on neural function. Chair of the Neurobiology Section. * Charles Stevens (Salk Institute): Synaptic physiology; physiological studies and biophysical models of synaptic plasticity in hippocampal neurons. * Roger Tsien (Chemistry): Second messenger systems in neurons; development of new optical and MRI probes of neuron function, including calcium indicators and caged neurotransmitters. * Mark Whitehead (Neurosurgery, School of Medicine): Peripheral and central taste systems; anatomical and functional studies of regions in the caudal brainstem important for feeding behavior. * Ruth Williams (Mathematics): Probabilistic analysis of stochastic systems and continuous learning algorithms. From Volker.Tresp at mchp.siemens.de Wed Dec 8 09:29:45 1999 From: Volker.Tresp at mchp.siemens.de (Volker Tresp) Date: Wed, 08 Dec 1999 15:29:45 +0100 Subject: Paper available on combining estimators and Gaussian process regression Message-ID: <384E6B59.DBB38C36@mchp.siemens.de> I would like to announce the availability of a new paper of potential interest to people working on combining estimators and Gaussian process regression or other kernel based systems. Comments are welcome! Greetings, - Volker ------------------------------------------------------------------------- A BAYESIAN COMMITTEE MACHINE Volker Tresp Siemens AG, Corporate Technology Neural Computation Dept. Information and Communications Otto-Hahn-Ring 6, 81730 Munich, Germany The Bayesian committee machine (BCM) is a novel approach to combining estimators which were trained on different data sets. Although the BCM can be applied to the combination of any kind of estimators the main foci are Gaussian process regression and related systems such as regularization networks and smoothing splines for which the degrees of freedom increase with the number of training data. Somewhat surprisingly, we find that the performance of the BCM improves if several test points are queried at the same time and is optimal if the number of test points is at least as large as the degrees of freedom of the estimator. The BCM also provides a new solution for online learning with potential applications to data mining. We apply the BCM to systems with fixed basis functions and discuss its relationship to Gaussian process regression. Finally, we also show how the ideas behind the BCM can be applied in a frequentist setting to extend the input dependent combination of estimators. ftp://flop.informatik.tu-muenchen.de/pub/hofmannr/bcm.ps.gz ----------------------------------------------------------------------- ----------------------------------------------------------------------- From sylee at ee.kaist.ac.kr Thu Dec 9 03:49:55 1999 From: sylee at ee.kaist.ac.kr (Soo-Young Lee) Date: Thu, 9 Dec 1999 17:49:55 +0900 Subject: Position Available : Research Professor and Post Doc Message-ID: <003d01bf4222$63b21640$329ef88f@kaist.ac.kr> I apologize for multiple receipants. The Brain Science Research Center (BSRC) and Department of Electrical and Computer Science at Korea Advanced Institute of Science and Technology have immediate openings for Research Professor and Post Doc. The Research Professor position requires at least 1 year experience after Ph.D. degree. The BSRC is the main research organization for Korean Brain Science and Engineering Research Program sponsored by Korean Ministry of Science and Technology. It currently manages a big national research program, and more than 100 professors throughout Korea are working in the research program. It is truly multidisciplinary research program, and ample opertunities exist to cooparate with researchers in other disciplines, from neuroscisnce to neural network applications and implementations. Although all aspects of biological and artificial neural neural networks are pursued, researchers in the following subjects have higher priority. (1) cognitive science (selective attention, etc.) (2) speech signal processing and recognition (3) auditory model (4) neuro-chip implementation (5) dynamic neural models WE EMPHASIZE INTERDISCIPLINARY RESEARCHES, FROM BIOLOGY TO HARDWARE THROUGH MODELING AND APPLICATIONS. Interested candidates should contact with Prof. Soo-Young Lee, Director, BSRC, at sylee at ee.kaist.ac.kr. From Sebastian_Thrun at heaven.learning.cs.cmu.edu Thu Dec 9 19:32:35 1999 From: Sebastian_Thrun at heaven.learning.cs.cmu.edu (Sebastian Thrun) Date: Thu, 09 Dec 1999 19:32:35 -0500 Subject: Call for papers: ICML-2k Message-ID: Please find enclosed the CFP for ICML-2000. It might interest you that ICML seeks to broaden its scope, and among other areas solicits submissions from the connectionists/statistical learning community. Consider submitting! It's a fun conference. -------------------------------------------------------------------------- Call for Papers THE SEVENTEENTH INTERNATIONAL CONFERENCE ON MACHINE LEARNING June 29-July 2, 2000 Stanford University The Seventeenth International Conference on Machine Learning (ICML-2000) will be held at Stanford University from June 29 to July 2, 2000, in the heart of Silicon Valley. The conference will bring together researchers to exchange ideas and report recent progress in the computational study of learning. Topics for Submission ICML-2000 welcomes submissions on all facets of machine learning, but especially solicits papers on problem areas, research topics, learning paradigms, and approaches to evaluation that have been rare at recent conferences, including: - the role of learning in natural language, vision and speech, planning and scheduling, design and configuration, logical and spatial reasoning, motor control, and more generally on learning for performance tasks carried out by intelligent agents; - the discovery of scientific laws and taxonomies, the construction of componential and structural models, and learning at multiple levels of temporal and spatial resolution; - the effect of the developers' decisions about problem formulation, representation, data quality, and reward function on the learning process; - computational models of human learning, applications to real-world problems, exploratory research that describes novel learning tasks, work that integrates familiar methods to demonstrate new functionality, and agent architectures in which learning plays a central role; - empirical studies that combine natural data (to show relevance) with synthetic data (to understand conditions on behavior), along with formal analyses that make contact with empirical results, especially where the aim is to identify sources of power, rather than to show one method is superior to others. Naturally, we also welcome submissions on traditional topics, ranging from induction over supervised data to learning from delayed rewards, but we hope the conference will also attract contributions on the issues above. Review Process The ICML-2000 review process will be structured to encourage publications covering a broad range of research and to foster increased participation in the conference. To this end, we have instituted: - area chairs who will be responsible for recruiting papers in their area of expertise and overseeing the review process for those submissions; - conditional acceptance of papers that are not publishable in their initial form, but that can be improved enough for inclusion in time to appear in the proceedings; and - a review form that requires referees to explicitly list any problems with a paper, what it would take to overcome them, and, if they recommend rejection, why it cannot be fixed in time for inclusion. The overall goal is to make the review process more like that in journals, with time for the authors to incorporate feedback from reviewers. Each submitted paper will be reviewed by two members of the program committee, with the decision about its acceptance overseen by the responsible area chair and the program chair. Paper Submission Authors should submit papers using same format and length as the final proceedings version. The detailed instructions for authors at http://www-csli.stanford.edu/icml2k/instructions.html include pointers to templates for LaTeX and Word documents. These specify two-column style, Times Roman font with 10 point type, vertical spacing of 11 points, overall text width of 6.75 inches, length of 9.0 inches, 0.25 inches between the two columns, top margin of 1.0 inch, and left margin of 0.75 inch. (The right and bottom margins will depend on whether one uses US letter or A4 paper.) Papers must not exceed eight (8) pages including figures and references. We will return to the authors any papers that do not satisfy these requirements. The deadline for submissions to ICML-2000 is MONDAY, JANUARY 24, 2000. Submission will be entirely electronic by transferring papers to the ICML-2000 ftp site, as explained in the detailed instructions for authors. Authors must submit papers in POSTSCRIPT format to ensure our ability to print them out for review. Each submission must be accompanied by the paper's title, the authors' names and physical addresses, a 250-word abstract, the contact author's email address and phone number, and the author who would present the talk at the conference. Authors must enter this information into the submission form at the conference web site by FRIDAY, JANUARY 21. ICML-2000 allows simultaneous submission to other conferences, provided this fact is clearly indicated on the submission form. Accepted papers will appear in the conference proceedings only if they are withdrawn from other conferences. Simultaneous submissions that are not clearly specified as such will be rejected. Other Conference Information The Seventeenth International Conference on Machine Learning will be collocated with the Thirteenth Annual Conference on Computational Learning Theory (COLT-2000) and the Sixteenth Conference on Uncertainty in Artificial Intelligence (UAI-2000). Registrants to any of these meetings will be able to attend the technical sessions of the others at no additional cost. ICML-2000 will also be preceded by tutorials on various facets of machine learning. For additional information, see the web site for the conference at http://www-csli.stanford.edu/icml2k/ which will provide additional details as they become available. If you have questions about ICML-2000, please send electronic mail to icml2k at csli.stanford.edu. The conference has received support from DaimlerChrysler Research and Technology, Stanford's Center for the Study of Language and Information (CSLI), and the Institute for the Study of Learning and Expertise (ISLE). From plong at comp.nus.edu.sg Thu Dec 9 23:11:02 1999 From: plong at comp.nus.edu.sg (Phil Long) Date: Fri, 10 Dec 1999 12:11:02 +0800 (GMT-8) Subject: computational learning theory postdoc Message-ID: The computational learning theory group of the National University of Singapore is looking for a postdoc. The focus of the project funding the position is the search for theoretically principled, practical algorithms for machine learning. The postdoc will be free to pursue independent research along these general lines, but will also have the opportunity to collaborate with the members of the group. The starting date is somewhat flexible, but should be some time before September, 2000. The position runs for two years. If you are interested, please send your CV and the names of three references to plong at comp.nus.edu.sg by January 15, 2000. From urbanczr at aston.ac.uk Fri Dec 10 09:18:55 1999 From: urbanczr at aston.ac.uk (R URBANCZIK) Date: Fri, 10 Dec 1999 14:18:55 +0000 Subject: paper on Imax Message-ID: The following paper (12 pages, submitted to Europhysics letters) is available online from: http://neural-server.aston.ac.uk/Papers/postscript/NCRG_1999_033.ps.Z Statistical Mechanics of Mutual Information Maximization R. Urbanczik, Neural Computing Research Group, Aston University Abstract: An unsupervised learning procedure based on maximizing the mutual information between the outputs of two networks receiving different but statistically dependent inputs is analyzed (Becker and Hinton, Nature, 355, 92, 161). By exploiting a formal analogy to supervised learning in parity machines, the theory of zero temperature Gibbs learning for the unsupervised procedure is presented for the case that the networks are perceptrons and for the case of fully connected committees. From rajeev at cns.bu.edu Fri Dec 10 18:22:19 1999 From: rajeev at cns.bu.edu (Rajeev Raizada) Date: Fri, 10 Dec 1999 18:22:19 -0500 (EST) Subject: Review of MIT CogSci Encyc downloadable Message-ID: Members of Connectionists might be interested in the following book review of "The MIT Encyclopedia of the Cognitive Sciences", forthcoming in the journal Neural Networks. A preprint is downloadable in PDF or PostScript formats at: http://cns-web.bu.edu/~rajeev/index.html#publications R. D. S. Raizada (1999). "A fruitful blend, or a trinket-box? A book review of The MIT Encyclopedia of the Cognitive Sciences." Forthcoming in Neural Networks. This review compares the MIT Encyclopedia to other recent reference works covering similar subject matter, e.g. M. S. Gazzaniga (Ed.) (1999) "The New Cognitive Neurosciences - 2nd Edition", and also discusses the extent to which it succeeds in showing interdisciplinary links between the various branches of Cognitive Science. Please let me know if you have any problems with downloading the preprint. Yours, Rajeev Raizada Dept. of Cognitive and Neural Systems Boston University 677 Beacon St. Boston, MA 02215. Tel: 617-353-6181 Fax: 617-353-7755 E.mail: rajeev at cns.bu.edu WWW: http://cns-web.bu.edu/~rajeev From F.Murtagh at Queens-Belfast.ac.uk Sat Dec 11 15:05:20 1999 From: F.Murtagh at Queens-Belfast.ac.uk (Prof. F. Murtagh) Date: Sat, 11 Dec 1999 20:05:20 +0000 (GMT) Subject: clustering and classification - _the_ source for bibliographies Message-ID: Dear Colleague, The following may be of interest, either to contribute a software or other exhibit area, or to get a copy when available (approx. May 2000). Regards, ------------------------------------------------------------- Prof. F. Murtagh, School of Computer Science, The Queen's University of Belfast, Belfast BT7 1NN, Northern Ireland Tel +44 28 9027-4620 Fax +44 28 9068-3890 http://www.cs.qub.ac.uk/~F.Murtagh f.murtagh at qub.ac.uk Centre for Image and Vision Systems http://www.qub.ac.uk/ivs ------------------------------------------------------------- Clustering and Classification Bibliographies - and now Software The Classification Literature Automated Search Service (CLASS), the only classification bibliography, published annually since the 1970s by the Classification Society of North America (CSNA), is undergoing big changes. From susanne at smi.auc.dk Mon Dec 13 04:48:30 1999 From: susanne at smi.auc.dk (Susanne Nielsen) Date: Mon, 13 Dec 1999 10:48:30 +0100 Subject: NEW DEADLINE FOR REGISTRATION AND SUBMISSION OF ABSTRACTS FOR IFESS 2000, NP 2000 WORKSHOP, AND PRE-CONFERENCE COURSE Message-ID: IFESS - YEAR 2000 - AALBORG, DENMARK ------------------------------------ IFESS 2000 CONFERENCE, 18TH-20TH JUNE NP 2000 WORKSHOP, 21ST-23RD JUNE PRE-CONFERENCE COURSE - MOTOR FUNCTIONS AND CNS INJURIES, 17TH-18TH JUNE NB: BECAUSE OF THE CHRISTMAS HOLIDAY WE HAVE POSTPONED THE DEADLINE FOR TWO WEEKS. NEW DEADLINE FOR REGISTRATION AND SUBMISSION OF ABSTRACTS IS THEREFORE 7TH JANUARY 2000 (POSTMARKED). http://www.smi.auc.dk/aalborg2000 1. The pre-conference course: "Motor Functions and CNS Injuries" - is designated as an international workshop to facilitate the communication between the various disciplines involved in the field of restoration of motor and sensory functions of individuals with motor disabilities - 17th-18th June 2000. See: http://www.smi.auc.dk/aalborg2000/course 2. The 5th Annual Conference of the International Functional Electrical Stimulation Society (IFESS 2000) - aims at bringing together internationally leading specialists in the field of Functional Electrical Stimulation (FES) to stimulate the exchange of views and new advancements and to discuss current and future goals in various applications of FES - 18th-20th June 2000. See: http://www.smi.auc.dk/aalborg2000/ifess 3. The Neural Prostheses VI: Motor Systems Conference (NP 2000) - will be organized as an international workshop to facilitate the communication between the various disciplines involved in the field of restoration of motor and sensory functions of individuals with motor disabilities - 21st-23th June 2000. See: http://www.smi.auc.dk/aalborg2000/np It is the first time in Europe that these events are organized simultaneously. The organizing Committee has aimed at having distinguished keynote and invited speakers to introduce all main topics, workshops, and sessions. We are expecting to have up to 200 participants from all over the world, among them about 50% graduate students and many clinicians. As a result of funding of the IFESS 2000, we now have grants available for Young Researchers. Further information can be found at: http://www.smi.auc.dk/aalborg2000/grants/grants.html The scientific programs for all three events, information on practical details, and registration form can be found at: http://www.smi.auc.dk/aalborg2000. If you have any further questions, please do not hesitate contacting us at: aalborg2000 at smi.auc.dk or at the following address: Center for Sensory-Motor Interaction (SMI) Aalborg University Fredrik Bajers Vej 7D-3 9220 Aalborg Denmark Phone: +45 96 35 88 28 Fax: +45 98 15 40 08 I take this oppotunity to wish you a merry Christmas and a happy New Year and do look forward to seeing you in Aalborg in the new millenium. Best regards, Thomas Sinkjaer Chairman Organising Committee IFESS 2000 & NP 2999 Professor, dr.med., Ph.D. --------------------------******--------------------------- Susanne Nielsen, Research Secretary Center for Sensory-Motor Interaction (SMI) Aalborg University Fredrik Bajers Vej 7D-3 9220 Aalborg Denmark Phone: +45 96 35 88 28 Fax: +45 98 15 40 08 E-mail: susanne at smi.auc.dk --------------------------******--------------------------- From psollich at mth.kcl.ac.uk Thu Dec 16 07:37:50 1999 From: psollich at mth.kcl.ac.uk (Peter Sollich) Date: Thu, 16 Dec 1999 12:37:50 +0000 (GMT) Subject: Postgraduate research open day Message-ID: Dear colleagues, on behalf of the Neural Networks and Disordered Systems group in the Department of Mathematics at King's College London, I am forwarding the announcement below of our forthcoming postgraduate open day. I would be grateful if you could encourage all interested (and suitably qualified) students to attend. Many thanks in advance for your help, Peter Sollich -------------------------------------------------------------------------- Peter Sollich Department of Mathematics Phone: +44 - (0)20 - 7848 2875 (new!) King's College Fax: +44 - (0)20 - 7848 2017 (new!) University of London E-mail: peter.sollich at kcl.ac.uk Strand WWW: http://www.mth.kcl.ac.uk/~psollich London WC2R 2LS, U.K. -------------------------------------------------------------------------- ***************************************************************************** * * * POSTGRADUATE OPEN DAY * * --------------------- * * * * The Mathematics Department of King's College London is holding an Open * * Day for prospective postgraduate students on the 18th of February, * * 2000. The day will include talks by members of the various research * * groups in the Department, as well as a panel discussion and small group * * discussions with prospective supervisors and current students. Coffee, * * lunch and tea will be provided, and the department will consider * * making a contribution towards students' travel expenses. * * * * King's College has one of the largest departments of mathematics in * * the United Kingdom and is one of the four London Colleges with the * * highest RAE research rating in both pure and applied mathematics. The * * mathematics department of King's is an exceptionally active, balanced * * and broad research unit, which offers research supervision in a wide * * range of subjects in pure and applied mathematics, including Analysis, * * Number Theory, Geometry and Topology, Theoretical Physics, General * * Relativity, * * NEURAL NETWORKS AND DISORDERED SYSTEMS, * * * * Statistical Mechanics, Quantum Probability, Mathematical Biology and * * Financial Mathematics. * * * * For full details of the Open Day programme, a registration form, and * * the booklet "Postgraduate Mathematics at King's", please contact * * * * The Postgraduate Secretary, Department of Mathematics, * * King's College London, Strand, London WC2R 2LS * * Phone 020-7848 2107, Fax 020-7848 2017, Email pgopenday.maths at kcl.ac.uk * * * * Further details and an electronic registration form are also available * * from the Department's web site at * * "http://www.mth.kcl.ac.uk/postgraduate/openday2000.html". * * * ***************************************************************************** From Roweis at gatsby.ucl.ac.uk Thu Dec 16 08:02:39 1999 From: Roweis at gatsby.ucl.ac.uk (Sam T. Roweis) Date: Thu, 16 Dec 1999 13:02:39 +0000 (GMT) Subject: NIPS tutorial notes on the web Message-ID: NIPS'99 Tutorial Notes Probabilistic Models for Unsupervised Learning Zoubin Ghahramani Sam Roweis We are making the slides from our recent tutorial available for download on the web. These slides correspond to the ones handed out in the NIPS tutorial booklet and also to those we used during the presentation. Feel free to distribute them as a package, but please let us know if you plan to include parts them in other documents. http://www.gatsby.ucl.ac.uk/~zoubin/NIPStutorial.html Enjoy the holidays, Sam & Zoubin ____________________________________________________________________________ Sam Roweis Gatsby Computational Neuroscience Unit, UCL 44 20 7 679 1188 voice 17 Queen Square, Room 503 44 20 7 679 1173 fax WC1N 3AR, London, UK roweis at gatsby.ucl.ac.uk http://www.gatsby.ucl.ac.uk/~roweis/ From eurich at theo.physik.uni-bremen.de Fri Dec 17 04:37:00 1999 From: eurich at theo.physik.uni-bremen.de (Christian Eurich) Date: Fri, 17 Dec 1999 10:37:00 +0100 Subject: Article on Hebbian learning Message-ID: <385A043C.93EB7A0C@physik.uni-bremen.de> Dear Connectionists, on the occasion of the recent NIPS workshop "Spike Timing and Synaptic Plasticity" organized by Paul Munro and Larry Abbott I would like to call your attention to the following article on the problem of delay adaptation and the stability of time sensitive Hebbian learning for different window functions. The article appeared in the Feb 15 issue of Physical Review Letters. Hardcopies are available on request. Christian W. Eurich, Klaus Pawelzik, Udo Ernst, Jack D. Cowan and John G. Milton, Dynamics of Self-Organized Delay Adaptation, Physical Review Letters 82 (1999) 1594-1597. Abstract: Adaptation of interaction delays is essential for the functioning of many natural and technical systems. We introduce a novel framework for studying the dynamics of delay adaptation in systems which optimize coincidence of inputs. For the important case of periodically modulated input we derive conditions for the existence and stability of solutions which constrain the set of mechanisms for reliable delay adaptation. Using numerical examples we show that our approach is applicable to more general than periodic input patterns such as Poissonian point processes with coordinated rate fluctuations. Cheers, Christian Eurich ================================================== Dr. Christian Eurich Universitaet Bremen Institut fuer Theoretische Neurophysik, FB 1 Postfach 330 440 D-28334 Bremen Germany Phone: +49 (421) 218-4559 Fax: +49 (421) 218-9104 Phone home: +49 (421) 5577208 Fax home: +49 (421) 5577050 e-mail: eurich at physik.uni-bremen.de homepage: http://www-neuro.physik.uni-bremen.de/~eurich "Die Welt ist alles, was der Fall ist." (L. W.) ================================================== From p.j.b.hancock at psych.stir.ac.uk Fri Dec 17 07:35:10 1999 From: p.j.b.hancock at psych.stir.ac.uk (Peter Hancock) Date: Fri, 17 Dec 1999 12:35:10 +0000 (GMT) Subject: Faculty posts in Computing Science Message-ID: The computing science department at the University of Stirling has vacancies for two chairs (full professors) and two lecturers. They would be very interested in people doing research in neural networks, to join the existing interdisciplinary group here. The official closing date has past, but anyone interested might still contact Prof. Ken Turner, kjt at cs.stir.ac.uk. http://www.cs.stir.ac.uk/jobs/ ------------------------------------------------------------------------- Peter Hancock Department of Psychology 0 0 Face Centre for University of Stirling | Research and Cognitive and FK9 4LA, UK \_/ Group Computational Phone 01786 467675 Fax 01786 467641 Neuroscience pjh at psych.stir.ac.uk http://www-psych.stir.ac.uk/~pjh ------------------------------------------------------------------------- From stefano.panzeri at ncl.ac.uk Fri Dec 17 09:23:19 1999 From: stefano.panzeri at ncl.ac.uk (Stefano Panzeri) Date: Fri, 17 Dec 1999 14:23:19 +0000 Subject: postdoctoral position in computational vision Message-ID: <385A4757.333556D8@ncl.ac.uk> Post-doctoral position in computational vision at the Neural Systems Group, Department of Psychology, University of Newcastle upon Tyne, UK http://www.psychology.ncl.ac.uk/neural_systems_group.html Applications are invited for a post-doctoral scientist position (up to ?30,626, depending on experience) for a research in computational neuroscience. The post is funded for 3 years in a first instance by an EU programme aimed at using concurrent psychophysics, computational models and neurophysiology to investigate mechanisms in visual neurons during natural image processing. Research topics for the computational neuroscience position will include: computational models of properties of visual neurons when they are stimulated with natural images; computational models of the role of prior knowledge of the statistics of the natural visual environment in visual perception; quantitative analysis of the response properties of visual neurons responding to natural images; developing new computational approaches for data in structure-function relationships (in particular relating connectivity, activation and information dynamics in the thalamo-cortical network). Applicants should possess a PhD in a relevant discipline (i.e. neuroscience, physics, computing, applied mathematics, engineering or others), a strong analytical background and a keen interest in vision. Experience and skills either in neural networks, information theory, visual psychophysics, visual neuroscience or computer image processing will be a great advantage. Departmental facilities for this work are excellent, and the research group has recently won more than $14M for the continuing improvement of its infrastructure. Please contact the Departmental Administrator for further details, at the Neural Systems Group, Department of Psychology, Ridley Building, Claremont Place, Newcastle upon Tyne, NE1 7RU. Tel: +44-191-222-7525, Fax: +44-191-222-5622. E-mail: m.p.young at ncl.ac.uk - From Nello.Cristianini at bristol.ac.uk Fri Dec 17 13:02:18 1999 From: Nello.Cristianini at bristol.ac.uk (N Cristianini) Date: Fri, 17 Dec 1999 18:02:18 +0000 (GMT) Subject: CFP: Support Vector - Special Issue Message-ID: Please post the following Call for Papers thanks Nello -------------------------------------------------------- Call for papers: ================ Special Issue of Machine Learning Journal on SUPPORT VECTOR MACHINES AND KERNEL METHODS Guest editors: Nello Cristianini (Bristol University, UK - contact person) Colin Campbell (Bristol University, UK) Chris Burges (Lucent Technologies, USA) Deadline: March 15th, 2000 More information on the webpage: http://lara.enm.bris.ac.uk/cig/specissue.htm or contact: Nello Cristianini, Department of Engineering Mathematics, Queen's Building, Bristol University, Bristol BS8 1TR United Kingdom Nello.Cristianini at bristol.ac.uk From lemm at lorentz.uni-muenster.de Mon Dec 20 13:29:13 1999 From: lemm at lorentz.uni-muenster.de (Joerg_Lemm) Date: Mon, 20 Dec 1999 19:29:13 +0100 (CET) Subject: Paper on Mixtures of Gaussian Process Priors Message-ID: Dear colleagues, The following paper (10 pages) is now available at http://xxx.lanl.gov/ps/physics/9911077 --------------------------------------------------------------------------- "Mixtures of Gaussian process priors" by Joerg Lemm, Muenster University Abstract: Nonparametric Bayesian approaches based on Gaussian processes have recently become popular in the empirical learning community. They encompass many classical methods of statistics, like Radial Basis Functions or various splines, and are technically convenient because Gaussian integrals can be calculated analytically. Restricting to Gaussian processes, however, forbids for example the implemention of genuine nonconcave priors. Mixtures of Gaussian process priors, on the other hand, allow the flexible implementation of complex and situation specific, also nonconcave "a priori" information. This is essential for tasks with, compared to their complexity, a small number of available training data. The paper concentrates on the formalism for Gaussian regression problems where prior mixture models provide a generalisation of classical quadratic, typically smoothness related, regularisation approaches being more flexible without having a much larger computational complexity. --------------------------------------------------------------------------- The paper and related material can also be found at my home page: http://pauli.uni-muenster.de/~lemm/ ======================================================================== Joerg Lemm Universitaet Muenster Email: lemm at uni-muenster.de Institut fuer Theoretische Physik I Phone: +49(251)83-34922 Wilhelm-Klemm-Str.9 Fax: +49(251)83-36328 D-48149 Muenster, Germany http://pauli.uni-muenster.de/~lemm ======================================================================== From harnad at coglit.ecs.soton.ac.uk Mon Dec 20 17:38:49 1999 From: harnad at coglit.ecs.soton.ac.uk (Stevan Harnad) Date: Mon, 20 Dec 1999 22:38:49 +0000 (GMT) Subject: AUTONOMOUS BRAIN/Milner: PSYC Call for Book Reviewers (673 lines) Message-ID: PSYCOLOQUY CALL FOR BOOK REVIEWERS Below is the Abstract of "The Autonomous Brain" by Peter M Milner This book has been selected for multiple review in Psycoloquy. If you wish to submit a formal book review please write to psyc at pucc.princeton.edu indicating what expertise you would bring to bear on reviewing the book if you were selected to review it. (If you have never reviewed for PSYCOLOQUY or Behavioral & Brain Sciences before, it would be helpful if you could also append a copy of your CV to your inquiry.) If you are selected as one of the reviewers and do not have a copy of the book, you will be sent a copy of the book directly by the publisher (please let us know if you have a copy already). Reviews may also be submitted without invitation, but all reviews will be refereed. The author will reply to all accepted reviews. FULL PSYCOLOQUY BOOK REVIEW INSTRUCTIONS AT: http://www.princeton.edu/~harnad/psyc.html http://www.cogsci.soton.ac.uk/psycoloquy/ Psycoloquy reviews are of the book not the Precis. Length should be about 200 lines [c. 1800 words], with a short abstract (about 50 words), an indexable title, and reviewer's full name and institutional address, email and Home Page URL. All references that are electronically accessible should also have URLs. AUTHOR'S RATIONALE FOR SOLICITING MULTIPLE REVIEW: "I hope this book will attract the attention of psychologists, philosophers and others interested in the mind, to some advances in neuroscience that are relevant to cognition. In particular, I think the idea that the neural processes involved in motivation and intention play an essential role in the learning of percepts should be more widely recognised. I acknowledge that no single person can hope to be well informed about the entire fields of neuroscience and cognition and it is my hope that multiple reviews of the work by experts may expose the inevitable inaccuracies and gaps, providing a partial substitute for multiple authorship. Having a number of reviews in one readily accessible place would provide a valuable resource for the reader." ----------------------------------------------------------------------- psycoloquy.99.10.071.autonomous-brain.1.milner Mon Dec 20 1999 ISSN 1055-0143 (50 paragraphs, 20 references, 620 lines) PSYCOLOQUY is sponsored by the American Psychological Association (APA) Copyright 1999 Peter M Milner THE AUTONOMOUS BRAIN [Lawrence Erlbaum Associates, Mahwah, NJ.155 pp. ISBN: 0-8058-3211-4] Precis of Milner on Autonomous-Brain Peter M. Milner McGill University Department of Psychology 1205 Dr. Penfield Ave. Montreal, QC. CANADA H3A 1B1 ps64 at musica.mcgill.ca ABSTRACT: This book presents a theory of behaviour based on the premise that nervous systems have evolved to enable animals to engage in a variety of useful activities. Information about the outside world is essential for most of these activities, but the theory that sensory input must shape its own information processing system is rejected. After the behaviourists banished the immaterial self from psychology they replaced it by stimulus input, and for many years behaviour was attributed entirely to sensory input. Only recently has the notion started to develop that it is the response mechanism of the brain that determines what stimuli are required to perform an action. In the model presented here, an executive system located in the frontal region of the brain employs the extensive reciprocal connections of the sensory cortex to select the input needed to guide the motor system. The consequences of having an autonomous response planner, instead of one subservient to outside stimuli, are far reaching. Neural representations of broad categories that can coexist with multiple distinct subclasses, and the related phenomenon of stimulus equivalence, become easier to understand, for example. It may also enable us to understand why we usually think that we make our own decisions. The book also has suggestions about the way serial order is learned and the role of the frontal regions of the brain in reinforcement, expectancy and response planning. KEYWORDS: association of ideas, attention, behaviour model, intention, motivation, self, serial order Full text of Precis of book available at: http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.071 or ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.071.autonomous-brain.1.milner From eurich at theo.physik.uni-bremen.de Tue Dec 21 08:43:29 1999 From: eurich at theo.physik.uni-bremen.de (Christian Eurich) Date: Tue, 21 Dec 1999 14:43:29 +0100 Subject: Preprints on the encoding accuracy of neural populations Message-ID: <385F8401.87B52DFA@physik.uni-bremen.de> Dear Connectionists, two preprints on Fisher information and the encoding accuracy of neural populations are available from my homepage at http://www-neuro.physik.uni-bremen.de/~eurich/Publications/publications.html 1. Multi-dimensional encoding strategy of spiking neurons C. W. Eurich and S. D. Wilke Neural Computation (in press) Abstract: Neural responses in sensory systems are typically triggered by a multitude of stimulus features. Using information theory, we study the encoding accuracy of a population of stochastically spiking neurons characterized by different tuning widths for the different features. The optimal encoding strategy for representing one feature most accurately consists of (i) narrow tuning in the dimension to be encoded to increase the single-neuron Fisher information, and (ii) broad tuning in all other dimensions to increase the number of active neurons. Extremely narrow tuning without sufficient receptive field overlap will severely worsen the coding. This implies the existence of an optimal tuning width for the feature to be encoded. Empirically, only a subset of all stimulus features will normally be accessible. In this case, relative encoding errors can be calculated which yield a criterion for the function of a neural population based on the measured tuning curves. 2. Neural representation of multi-dimensional stimuli C. W. Eurich, S. D. Wilke and H. Schwegler Proc. NIPS '99 (in press). Abstract: The encoding accuracy of a population of stochastically spiking neurons is studied for different distributions of their tuning widths. The situation of identical radially symmetric receptive fields for all neurons, which is usually considered in the literature, turns out to be disadvantageous from an information-theoretic point of view. Both a variability of tuning widths and a fragmentation of the neural population into specialized subpopulations improve the encoding accuracy. Both preprints are available in postscript and g-zipped postscript format. ================================================== Dr. Christian Eurich Universitaet Bremen Institut fuer Theoretische Neurophysik, FB 1 Postfach 330 440 D-28334 Bremen Germany Phone: +49 (421) 218-4559 Fax: +49 (421) 218-9104 Phone home: +49 (421) 5577208 Fax home: +49 (421) 5577050 e-mail: eurich at physik.uni-bremen.de homepage: http://www-neuro.physik.uni-bremen.de/~eurich "Die Welt ist alles, was der Fall ist." (L. W.) ================================================== From smagt at dlr.de Tue Dec 21 10:18:28 1999 From: smagt at dlr.de (Patrick van der Smagt) Date: Tue, 21 Dec 1999 16:18:28 +0100 (MET) Subject: CFP: Scalable Robotic Applications of Neural Networks Message-ID: <199912211518.QAA18903@ilz.robotic.dlr.de> From smagt at dlr.de Tue Dec 21 10:18:28 1999 From: smagt at dlr.de (Patrick van der Smagt) Date: Tue, 21 Dec 1999 16:18:28 +0100 (MET) Subject: CFP: Scalable Robotic Applications of Neural Networks Message-ID: <199912211518.QAA18903@ilz.robotic.dlr.de> REMINDER: CALL FOR PAPERS Scalable Robotic Applications of Neural Networks http://www.robotic.dlr.de/Smagt/CFP/ Special Issue of "Applied Intelligence" Submission deadline: January 15, 2000 In this special issue, we want to address the question of whether there are real robotic problems that can be better solved using existing neuro-computational principles than by using other standard engineering techniques. Where sensory-motor systems cannot be explicitly modeled, the brain's success leads us to expect that approaches based on adaptive neural control will someday provide a technically sound alternative. However, today's robotic engineers are appropriately skeptical regarding use of neural network principles because of the apparent scarcity of published research that demonstrates scalability of solutions to complex robotic control tasks. For example, like many traditional approaches, many neural network control strategies do not scale well from a two-degree-of-freedom robot arm to a seven degree-of-freedom system. For this special issue, papers are sought that: 1. introduce or review biologically plausible models of sensory-motor control that truly integrate action and perception; 2. newly apply such models to the control of realistic robots, especially manipulators; 3. describe hybrid robot control methodologies that incorporate cerebellar or other neuro-computational models (e.g., vision); 4. provide a case history that clearly defines, or illustrates overcoming of, barriers to successful competition by neural network models for robot control. The scalability of applications of biological models can only be fully assessed in the context of demonstrations that are representative of real-world complexities. Therefore, results on real hardware will generally be preferred. However, results on well-simulated complex problems will be preferred to results with hardware (e.g., 2 DoF robot arms) that can already be optimally controlled with conventional, non-neural techniques. -------------------------------------- Special Issue Editors Patrick van der Smagt Institute of Robotics and System Dynamics DLR (German Aerospace Research Center) Oberpfaffenhofen, Germany smagt at dlr.de http://www.robotic.dlr.de/Smagt/ Daniel Bullock Cognitive and Neural Systems Department Boston University danb at cns.bu.edu http://cns-web.bu.edu/faculty.html#bullock Editor in Chief Moonis Ali Department of Computer Science Southwest Texas State University ma04 at swt.edu http://www.cs.swt.edu -------------------------------------- Submission Deadline for submitting papers: Jan 15, 2000. The journal's instructions to authors can be read here. For this special issue, for review all papers must be submitted by the above date in Postscript or PDF format via email to smagt at dlr.de. Hardcopy submissions will not be accepted. To the extent that time permits, the editors will invite short external commentaries (to be published in the same issue) on the accepted papers. Submitters should inform the editors if they do *not* want their paper treated in an invited commentary. Suggestions of expert reviewers and commentators are welcome. -------------------------------------- About the journal The objective of Applied Intelligence (IJAI) is to provide a medium for exchanging applied research on intelligent systems and technological achievements. IJAI is currently in its eighth year of publication. In order to meet the demand, the publication frequency has been increased from four to six issues per year effective 1998. IJAI is abstracted and/or indexed by twenty indexing publications. See http://www.wkap.nl/journals/apin. Wishing you all a Merry Christmas and a Happy New Year! Patrick -- Dr Patrick van der Smagt phone +49 8153 281152, fax -34 DLR/Institute of Robotics and Mechatronics smagt at dlr.de P.O.Box 1116, 82230 Wessling, Germany http://www.robotic.de/Smagt/ From stork at rsv.ricoh.com Tue Dec 21 17:18:09 1999 From: stork at rsv.ricoh.com (David G. Stork) Date: Tue, 21 Dec 1999 14:18:09 -0800 (PST) Subject: Open Mind handwriting project Message-ID: ======================================================================== [[Distribute freely]] *** *** *** Request for assistance with Open Mind Handwriting Recognition *** *** *** Led by Dr. Lambert Schomaker of the UNIPEN Foundation (www.UNIPEN.org), the Open Mind Handwriting Recognition project seeks to collect data and develop recognizers of a variety of handwriting data, such as digits, isolated letters, and words in a variety of western languages (primarily English). Handwriting Recognition requires large sets of training data (as do other Open Mind projects) and our hypothesis is that such data can be contributed by non-expert netizens over the web. You can read background material at: www.OpenMind.org/OpenMindICDAR.pdf and www.OpenMind.org/OpenMindICDAR.typeset.pdf In a character-based approach, images of unknown recorded isolated character patterns are presented on a netizen's browser. The netizen then submits to the Open Mind host his or her classification of the character by clicking the corresponding button on a simple interface. For the case of words (having a known ASCII string representation but an unknown location of the letters in the image), the interface is a bit more complex but the general idea remains the same: humans are quite good at pointing out the individual characters in a word. While a small-scale web-based labeling system has been written, there is much software engineering to be done before it is "ready for prime time." Several Linux example programs for labeling are available to serve as an inspiration for developing the browser version for netizens. Issues of browser independence, speed, load optimizing, robustness, algorithms for ensuring data reliability, and the balancing of client/server computation must be addressed before the project is ready for load testing and public deployment. We are seeking software engineers to contribute gratis to this effort. Practical experience with Web-based user interfaces, Java, Javascript, CGI programming and Linux (C language) will be required. A pragmatic view, combined with sensitivity to user issues, would be ideal. A knowledge of pattern recognition research, while desirable, is not needed. If you feel you would like to contribute, please send a short e-mail describing your relevant background to me (not Dr. Schomaker directly). Of course, if you have particular questions, please don't hesitate to contact me. In either case, please write "Open Mind Handwriting" in your header. Many thanks in advance. --David Stork stork at OpenMind.org ======================================================================== | The Open Mind Initiative (www.OpenMind.org) is a collaborative | | framework for developing 'intelligent' software over the internet. | | Open Mind extends traditional open source methodology to allow | | non-expert 'netizens' to contribute data needed for such software. | | | | | | To subscribe to a moderated, low-traffic mailing list on Open Mind, | | send email to majordomo at OpenMind.org with the following message: | | subscribe openmind-general | | | | To be removed from the list, send instead the message: | | unsubscribe openmind-general | ======================================================================== From mtipping at microsoft.com Thu Dec 23 12:41:39 1999 From: mtipping at microsoft.com (Michael Tipping) Date: Thu, 23 Dec 1999 09:41:39 -0800 Subject: Paper announcement: the Relevance Vector Machine Message-ID: <2DCBFADAFCBBD21189D400805F6FA1AB0D334209@RED-MSG-12> Dear Connectionists, A paper on the "relevance vector machine", a Bayesian implementation of sparse kernel regression and classification models (which is to appear in the forthcoming proceedings of NIPS*12), is now available online: ftp://ftp.research.microsoft.com/users/mtipping/rvm_nips.ps.gz Title and abstract are below: -------------------------------------------------------------------------- The Relevance Vector Machine Michael E. Tipping The support vector machine (SVM) is a state-of-the-art technique for regression and classification, combining excellent generalisation properties with a sparse kernel representation. However, it does suffer from a number of disadvantages, notably the absence of probabilistic outputs, the requirement to estimate a trade-off parameter and the need to utilise 'Mercer' kernel functions. In this paper we introduce the Relevance Vector Machine (RVM), a Bayesian treatment of a generalised linear model of identical functional form to the SVM. The RVM suffers from none of the above disadvantages, and examples demonstrate that for comparable generalisation performance, the RVM requires dramatically fewer kernel functions. From steve at cns.bu.edu Fri Dec 24 11:09:52 1999 From: steve at cns.bu.edu (Stephen Grossberg) Date: Fri, 24 Dec 1999 11:09:52 -0500 Subject: neural dynamics of speech perception and word recognition Message-ID: The following article is available at http://www.cns.bu.edu/Profiles/Grossberg in PDF and G'zipped Postscript versions. Grossberg, S. and Myers, C.W. (2000). The resonant dynamics of speech perception: Interword integration and duration-dependent backward effects. Psychological Review, in press. ABSTRACT: How do listeners integrate temporally distributed phonemic information into coherent representations of syllables and words? During fluent speech perception, variations in the durations of speech sounds and silent pauses can produce different perceived groupings. For example, increasing the silence interval between the words "gray chip" may result in the percept "great chip", whereas increasing the duration of fricative noise in "chip" may alter the percept to "great ship" (Repp et al., 1978). The ARTWORD neural model quantitatively simulates such context-sensitive speech data. In ARTWORD, sequential activation and storage of phonemic items in working memory provides bottom-up input to unitized representations, or list chunks, that group together sequences of items of variable length. The list chunks compete with each other as they dynamically integrate this bottom-up information. The winning groupings feed back to provide top-down support to their phonemic items. Feedback stablishes a resonance which temporarily boosts the activation levels of selected items and chunks, thereby creating an emergent conscious percept. Because the resonance evolves more slowly than working memory activation, it can be influenced by information presented after relatively long intervening silence intervals. The same phonemic input can hereby yield different groupings depending on its arrival time. Processes of resonant transfer and competitive teaming help determine which groupings win the competition. Habituating levels of neurotransmitter along the pathways that sustain the resonant feedback lead to a resonant collapse that permits the formation of subsequent resonances. From fayyad at MICROSOFT.com Sat Dec 25 02:55:30 1999 From: fayyad at MICROSOFT.com (Usama Fayyad) Date: Fri, 24 Dec 1999 23:55:30 -0800 Subject: SIGKDD Explorations: vol. 1:2, announcements, news, and advertise ments Message-ID: The first issue of SIGKDD Explorations, the official newsletter of the ACM's new Special Interest Group (SIG) on Knowledge Discovery and Data Mining was published in June 1999 and was sent to all SIGKDD members and ACM network of libraries. It will also be included in SIGMOD's DiSC anthology CD-ROM. If you are a member and did not receive your hardcopy, please e-mail: help at acm.org to fix any problems with your membership. The first vlume is available to all on the web at: http://research.microsoft.com/datamine/SIGKDD. If you have any short news items or announcements approriate for inclusion in the second issue, please submit them to fayyad at acm.org or sunita at cs.berkeley.edu. For more information and examples from issue 1, please see: http://research.microsoft.com/datamine/SIGKDD. SIGKDD Explorations is a twice-annual newsletter dedicated to serve the SIGKDD membership and community. Anoouncements should not exceed 4 lines in length. Please provide URL if there is associated web information. *NEW*: We also have added a for-pay advertisement section to allow vendors, companies, consultants, and others to reach the rapidly growing SIGKDD community. Advertising rates start at $250 for quarter page, $500 per half page, and $1000 for a full page. All submissions must arrive by Jan. 20, 1999 for inclusion in the next issue. thanks, Usama Fayyad, Editor-in-Chief (fayyad at acm.org) Sunita Sarawagi, Associate Editor (sunita at cs.berkeley.edu) http://research.microsoft.com/datamine/SIGKDD From doya at ctr.atr.co.jp Sat Dec 25 06:13:21 1999 From: doya at ctr.atr.co.jp (Kenji Doya) Date: Sat, 25 Dec 1999 20:13:21 +0900 Subject: Research positions: computational neuroscience and robotics Message-ID: <3864A5BE.F93732E1@ctr.atr.co.jp> Post-doctoral Research Positions METALEARNING, NEUROMODULATION AND EMOTION http://meta.rad.atr.co.jp/ CREST, Japan Science and Technology Corporation The goal of the research project is to understand the mechanisms of higher-level regulation of learning functions of the brain. We propose computational models of the roles of the neuromodulators (e.g., dopamine, serotonin, noradrenaline, acetylecholine), test them by neurobiological and clinical experiments, and explore the functions of emotion-like systems in adaptive communication agents. Post-doctoral research positions are available immediately for the studies of 1) modeling the functions of neuromodulators in close collaboration with the experimental laboratories below; 2) development of self regulated learning algorithms for adaptive agents; and 3) designing and building a colony of robots with rich sensory and motor functions. Candidates should have a strong background in theoretical and/or experimental sciences, including neurobiology, psychology, computer science, robotics, physics and mathematics. Competitive salaries and benefits are provided by JST (Japan Science and Technology Corporation). The laboratory is located at ATR International in Kyoto, where people from diverse origins and backgrounds get together for basic research on human cognition and telecommunication. The research project is supported by JST's "Creating the Brain" program, supervised by Dr. Shun-ichi Amari. The principal project members and their research topics are the following: Takashi Matsumoto (Waseda University): Development of on-line Bayesian learning theory and algorithms. Shin Ishii (Nara Institute of Science and Technology): Development of metalearning algorithms and architectures for adaptive agents. Yuko Sekino (Gunma University): In vivo and in vitro experiments on regulation of hippocampal memory by neuromodulators in rats and mice. Toshiyuki Sawaguchi (Hokkaido University): Experiments on the roles of neuromodulators in the prefrontal cortex in monkeys. Shigeto Yamawaki (Hiroshima University): Behavioral and pharmacological studies on the roles of serotonin in human cognition using fMRI, MEG and TMS. Kenji Doya (ATR International): Development of a computational theory of neuromodulator functions; Real and simulated experiments of learning and communication in multiple agents. Applicants should send a CV, a statement of research interests, and recent reprints and/or preprints by January 31st, 2000 to Kenji Doya ATR International 2-2 Hikaridai, Seika, Soraku Kyoto 619-0288, Japan For more information, please contact info at meta.rad.atr.co.jp or visit our web page at http://meta.rad.atr.co.jp/ -- Kenji Doya ATR International 2-2 Hikaridai, Seika, Soraku, Kyoto 619-0288, Japan Phone:+81-774-95-1251, Fax:+81-774-95-1259 E-mail:doya at ctr.atr.co.jp, http://meta.rad.atr.co.jp/doya/