From iiass.alfredo at tin.it Mon Jul 3 13:26:43 2000 From: iiass.alfredo at tin.it (Alfredo Petrosino) Date: Mon, 03 Jul 2000 19:26:43 +0200 Subject: Neural Nets School 2000 Message-ID: <3960CCD3.65CD2D10@tin.it> 5th Course of International Summer School "Neural Nets E. R. Caianiello" on Visual Attention Mechanisms 23-29 October 2000 International Institute for Advanced Scientific Studies (IIASS) Vietri sul Mare, Salerno (Italy) http://www.iiass.it/nnschool JOINTLY ORGANIZED BY International Institute for Advanced Scientific Studies (IIASS) Ettore Majorana Foundation and Center for Scientific Culture (EMFCSC) SPONSORED BY Salerno University, Department of "Scienze Fisiche", Italy Gruppo Nazionale di Cibernetica e Biofisica del CNR Pavia University, Department of Electrical Engineering, Italy DIRECTOR OF THE 5TH COURSE Virginio CANTONI (Pavia University, Italy) DIRECTORS OF THE SCHOOL Michael JORDAN (University of California, Berkely, USA) Maria MARINARO (Salerno University, Italy) ORGANIZING COMMITEE Virginio CANTONI (Pavia University, Italy) Maria MARINARO (Salerno University, Italy) Alfredo PETROSINO (INFM-Salerno University, Italy) The school, open to all suitably qualified scientists from around the world, is organized in lectures, panel discussions and poster presentations and will cover a number of broad themes relevant to Visual Attention, among them: - Foundation: Early vision, Visual streams, Perception and action, Log-map analysis - Attentional mechanisms: Pop-out theory, Texton theory, Contour integration and closure, Fuzzy engagement mechanisms - Visual search : Attentional control, Selective attention, Spatial attention, Detection versus discrimination - Multiresolution and planning : Complexity of search tasks, Hierarchical perceptual loops, Multiresolution and associative memory systems, Attention and action planning - Attentional Visual Architectures: Neural models of visual attention, Hierarchical and associative networks, Attentional pyramidal neural mechanisms - Experiences: Eyeputer and scanpath recorders, etc. INVITED SPEAKERS (the list is not complete): Virginio CANTONI, Pavia University Leonardo CHELAZZI, Verona University, Italy Vito DI GESU`, Palermo University, Italy Hezy YESHURUN, Haifa University, Israel Zhaoping LI, Gatsby, University College, London, UK Luca LOMBARDI, Pavia University, Italy Carlo Alberto MARZI, Verona University, Italy Alain MERIGOT, University of Paris Sud, France Eliano PESSA, Roma University, Italy Alfredo PETROSINO, INFM-Salerno University, Italy Marco PIASTRA, Pavia University, Italy Vito ROBERTO, Udine University, Italy Dov SAGI, Weizmann University, Israel John TSOTSOS, Center for Computer Vision, Canada Daniela ZAMBARBIERI, Pavia University, Italy Harry WECHSLER, George Mason University, USA Steven YANTIS, Johns Hopkins University, USA POETIC TOUCH Vietri (from "Veteri", its ancient Roman name) sul Mare ("on sea") is located within walking distance from Salerno and marks the beginning of the Amalfi coast. Short rides take to Positano, Sorrento, Pompei, Herculaneum, Paestum, Vesuvius, or by boat, the islands of Capri, Ischia, and Procida. Velia (the ancient "Elea" of Zeno and Parmenide) is a hundred kilometers farther down along the coast. Visit http://www.comune.vietri-sul-mare.sa.it/ for more information. FEE The full school fee is 1200 dollars, reduced to 1000 dollars for students. The fee includes accommodations in twin room, meals, one day of excursion, and a copy of the proceedings of the school. A supplement of 40 dollars per night should be paid for single room. A restricted number of scholarships are also available for students which make the specific request. DATES Application deadline: 8 September 2000 Acceptance notification : 20 September 2000 For further information please contact : Dr. A.Petrosino Fax: + 39 89 761189 Email: iiass.alfredo at tin.it ============================CUT====================================== APPLICATION FORM Title:_______ Family Name: ________________________________________________________ Other Names:_________________________________________________________ Name to appear on badge: ____________________________________________ MAILING ADDRESS: Institution _________________________________________________________ Department __________________________________________________________ Address _____________________________________________________________ State ____________________________ Country __________________________ Phone:____________________________ Fax: _____________________________ E-mail: _____________________________________________________________ Arrival date: __________________ Departure date: ____________________ Will you be applying for a scholarship ? yes/no (Please include in your application the amount of bursary support and a justification for the request) Will you submit a poster ? yes/no (Please include a one page abstract for review by the organizers). ============================CUT====================================== Please send the application form by electronic mail to: iiass.alfredo at tin.it, subject: Neural Nets school; or by fax to: Neural Nets School, +39 89 761 189 or by ordinary mail to the address: Neural Nets School IIASS, Via Pellegrino 19 I84019 Vietri sul Mare (Sa) Italy From fritz at neuro.informatik.uni-ulm.de Tue Jul 4 09:15:52 2000 From: fritz at neuro.informatik.uni-ulm.de (Fritz Sommer) Date: Tue, 4 Jul 2000 15:15:52 +0200 (MET DST) Subject: No subject Message-ID: <200007041315.PAA07533@cerebellum.informatik.uni-ulm.de> cneuro at bbb.caltech.edu, bisc-group at cs.berkely.edu From kap-listman at wkap.nl Tue Jul 4 20:07:19 2000 From: kap-listman at wkap.nl (kap-listman@wkap.nl) Date: Wed, 05 Jul 2000 02:07:19 +0200 (MEST) Subject: New Issue: Neural Processing Letters. Vol. 11, Issue 3 Message-ID: <200007050007.CAA27310@wkap.nl> Kluwer ALERT, the free notification service from Kluwer Academic/PLENUM Publishers and Kluwer Law International ------------------------------------------------------------ Neural Processing Letters ISSN 1370-4621 http://www.wkap.nl/issuetoc.htm/1370-4621+11+3+2000 Vol. 11, Issue 3, June 2000. TITLE: A Batch Learning Vector Quantization Algorithm for Nearest Neighbour Classification AUTHOR(S): Sergio Bermejo, Joan Cabestany KEYWORD(S): Learning Vector Quantization, Newton-s optimization, nearest neighbour classification, batch learning algorithms. PAGE(S): 173-184 TITLE: A Comparison between the Tikhonov and the Bayesian Approaches to Calculate Regularisation Matrices AUTHOR(S): Andreu Catala, Cecilio Angulo KEYWORD(S): Bayesian inference, ill-posed problems, neural networks, RBF, regularization techniques, smoothing functions. PAGE(S): 185-195 TITLE: Competitive and Temporal Inhibition Structures with Spiking Neurons AUTHOR(S): E. Ros, F. J. Pelayo, P. Martin-Smith, I. Rojas, D. Palomar, A. Prieto KEYWORD(S): spiking neurons, competitive processing, temporal inhibition, attentional control mechanisms, bio-inspired neural systems. PAGE(S): 197-208 TITLE: An Experimental Comparison of Three PCA Neural Networks AUTHOR(S): Simone Fiori KEYWORD(S): principal component analysis, generalized Hebbian learning, adaptive principal-component extraction. PAGE(S): 209-218 TITLE: Neural Net Based Hybrid Modeling of the Methanol Synthesis Process AUTHOR(S): Primoz Potocnik, Igor Grabec, Marko Setinc, Janez Levec KEYWORD(S): hybrid modeling, genetic algorithms, feature selection, methanol synthesis, neural networks. PAGE(S): 219-228 TITLE: Transfusion Cost Containment for Abdominal Surgery with Neural Networks AUTHOR(S): Steven Walczak, John E. Scharf KEYWORD(S): neural networks, abdominal surgery, AAA, transfusion, cost, MSBOS. PAGE(S): 229-238 -------------------------------------------------------------- Thank you for your interest in Kluwer's books and journals. NORTH, CENTRAL AND SOUTH AMERICA Kluwer Academic Publishers Order Department, PO Box 358 Accord Station, Hingham, MA 02018-0358 USA Telephone (781) 871-6600 Fax (781) 681-9045 E-Mail: kluwer at wkap.com Kluwer Law International Order Department 675 Massachusetts Avenue Cambridge, MA 02139 USA Telephone: (617) 354-0140 Toll-free (US customers only): 800 577-8118 Fax: (617) 354-8595 E-mail: sales at kluwerlaw.com EUROPE, ASIA AND AFRICA Kluwer Academic Publishers Distribution Center PO Box 322 3300 AH Dordrecht The Netherlands Telephone 31-78-6392392 Fax 31-78-6546474 E-Mail: orderdept at wkap.nl From rfrench at ulg.ac.be Wed Jul 5 08:50:24 2000 From: rfrench at ulg.ac.be (Robert French) Date: Wed, 05 Jul 2000 14:50:24 +0200 Subject: POSTDOCTORAL AND Ph.D OPPORTUNITIES FOR EUROPEANS in connectionist neural network modelling Message-ID: <4.1.20000705133723.00ad24c0@pop3.mailst.ulg.ac.be> POSTDOCTORAL AND Ph.D OPPORTUNITIES FOR EUROPEANS in connectionist neural network modelling We have received a four-year grant from the European Commission grant to study the basic mechanisms of learning and forgetting in natural and artificial neural systems. The work will be done at five universities in England, Belgium and France and will be a highly multi-disciplinary effort involving experimental psychology, computer and mathematical modelling, and neural imaging techniques. There is funding for 5 fixed-term (1 to 2 years) post-doctoral research fellows, and 5 PhD studentships distributed across the participating institutions. Learning new information can potentially interfere severely with previously learned information. In order to prevent this from happening, many researchers now believe that the brain evolved a =93dual memory=94 architecture in which information is first processed in the hippocampus and thereafter gradually consolidated in the neo-cortex. But exactly how this works is still far from clear. We hope to gain a better understanding of the mechanisms and the implications of hippocampal-neocortical information transfer. Beyond developing a formal understanding of this system, the project will also explore applications of this type of memory architecture in the areas of child development, implicit learning, and various forms of aphasia and amnesia. This research has a number of important implications for understanding human memory. In particular, the dual-network connectionist architecture makes a number of rather unexpected predictions concerning the evolution over time of representations in long-term memory. One of the major goals of this project is to determine whether or not this representational evolution occurs in the brain. This work may also shed light on infant memory consolidation and category-specific deficits observed in certain types of aphasia. The project includes funding for 5 fixed-term (1 to 2 years) post-doctoral research fellows, and 5 PhD studentships. The focus of research for these positions will depend on the institution of appointment. Although there is considerable overlap between the expertise in the different partner institutions, in general, the research focus at each institution will be as follows: University of Li=E8ge, Belgium (with Bob French): investigations of memory consolidation and forgetting using experimental and modelling methodologies University of Grenoble, France (Bernard Ans/St=E9phane Rousset): investigations of transfer mechanisms between network systems and between hippocampus and neocortex using modeling and neural imaging methodologies University of Warwick, UK (Nick Chater): Formal analysis of data compression during information transfer between network systems Birkbeck College, UK (Denis Mareschal): implications of dual memory system for understanding infant and cognitive development using experimental and modelling methodologies Universit=E9 Libre de Bruxelles, Belgium (Axel Cleeremans): investigations of implicit learning and consciousness using experimental and modelling methodologies The program is highly multi-disciplinary with a particular emphasis on connectionist modelling. The members of the project will do experimental work as well as modelling and will be based in one of the five universities but will be expected to interact at a high level with the other teams involved in the project. For example, a student focusing on modelling at the University of Li=E8ge will interact with and visit the University of Grenoble to acquire neural imaging skills. Considerable emphasis will be placed on an exchange of ideas among project participants and publication in international journals. The ideal candidate will have good programming skills, a knowledge of connectionist models, experimental skills, an interest in human memory, and a willingness to work in an inter-disciplinary setting. The candidate must be aged 35 years or less at the time of appointment (excluding time spent for compulsory military service or childcare). They must be nationals of a European Community Member State or an Associated State* or must have resided in the community for the last 5 years prior to their appointment. They must not be nationals of the host country they are applying to and must not have carried out their normal activities in the host country for more than 12 of the 24 months prior to their appointment. Doctoral candidates must have completed their undergraduate degree and post-doctoral candidates must have completed their doctorate. Knowledge of the language of the host country is helpful, but is not necessary. The language of the project is English. Salaries for these positions vary but are competitive. The exact amounts are stipulated by the regulations of the participating universities. For more information on the exact value of the stipends, please contact the project director at the university that you are interested in attending. These positions are available immediately. There is no closing deadline for application. Applications will continue to be accepted until the positions are filled. Interested candidates should directly contact the following individuals at the participating institutions for further details: Bob French (rfrench at ulg.ac.be) at the University of Liege Denis Mareschal (d.mareschal at bbk.ac.uk) at Birkbeck College Nick Chater (nick.chater at warwick.ac.uk) at Warwick University Axel Cleeremans (axcleer at ulb.ac.be) at the Universite Libre de Bruxelles Bernard Ans (Bernard.Ans at upmf-grenoble.fr) at the University of Grenoble. If you have any questions concerning the project, feel free to contact the project director, Bob French (rfrench at ulg.ac.be). *European Union Associated States are: Bulgaria, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, Slovenia, Lichtenstein, Iceland, Israel, and Norway. ---------------------------------------------------------------------------- Robert M. French, Ph.D Quantitative Psychology and Cognitive Science Psychology Department University of Liege 4000 Liege, Belgium Tel: (32.[0]4) 366.20.10 FAX: (32.[0]4) 366.28.59 email: rfrench at ulg.ac.be URL: http://www.fapse.ulg.ac.be/Lab/cogsci/rfrench.html ---------------------------------------------------------------------------- From edwin at arti.vub.ac.be Wed Jul 5 12:58:18 2000 From: edwin at arti.vub.ac.be (Edwin de Jong) Date: Wed, 5 Jul 2000 18:58:18 +0200 (MET DST) Subject: PhD thesis available: Autonomous Formation of Concepts and Communication Message-ID: <200007051658.SAA07270@arti6.vub.ac.be> Dear Connectionists, this is to announce the availability of my PhD thesis for download. URL: http://arti.vub.ac.be/~edwin/thesis Title: Autonomous Formation of Concepts and Communication Best regards, Edwin de Jong __ Keywords: concept formation, evaluative feedback, reinforcement learning, situation concepts, development of communication, dynamical systems. Abstract Autonomous agents receive sensor input from the environment, select actions, and may receive evaluative feedback on their behavior. Multiple agents that are present in the same environment can benefit from using communication to overcome uncertainty in their information about the environment. The research in the thesis addresses the question of how concepts about the environment can be formed, and how a system of communication can develop that allows agents to exchange information about their environment. Instead of assuming that concepts are already available or based on universal primitives, in the approach followed here they are formed in response to interaction with the environment. Evaluative feedback can play an important role in this. This distinguishes the approach from concept learning, which requires supervised feedback. A particular type of concepts is described, called situation concepts. Situation concepts consist of features in the history of interaction between the agent and its environment, and predict some aspect of the future evolution of the state of the environment, possibly conditioned on the actions the agent may take. Several existing methods, particularly from the field of reinforcement learning, can be viewed as constructing a form of situation concepts, and a particular method for constructing a specific type of situation concepts is described. Since situation concepts convey information about the environment, they are especially suited for use in communication. The development of communication consists of the formation of associations between words and the concepts formed by individual agents, and is viewed as the behavior of a dynamical system. The variables of this system are the strengths of the associations between words and the concepts of all agents in a population. An algorithm for association formation for individual agents is described that leads to a common system of communication. A deterministic version of the system is shown mathematically and demonstrated experimentally to have point attractors that correspond to perfect communication. The stochastic system system has points in its phase space that play a similar role, and is preferable since it avoids certain deadlock situations. This finding is confirmed by an investigation of the relation between the amount of stochasticity and the quality of communication. The research contributes to the view of the development of communication as the behavior of a dynamical system. Finally, systematic measures are provided for the quality of conceptual systems and communication systems that can be used when the subject of communication, called referent, is known. From sheri.mizumori at psych.utah.edu Wed Jul 5 17:54:06 2000 From: sheri.mizumori at psych.utah.edu (Sheri Mizumori) Date: Wed, 05 Jul 2000 15:54:06 -0600 Subject: behavioral neuroscience postdoc positions at the University of Washington Message-ID: <3963AE7E.77C4FDC8@csbs.utah.edu> Post Doctoral Positions in Behavioral Neuroscience Two NIH funded positions are available for research concerning the neurobiology of learning and memory. Recent research from this lab has investigated the relative contributions of multiple brain structures to adaptive navigation by rats. These and other ongoing studies involve parallel recordings of single unit activity within single brain as well as across diverse neural systems. Our goal is to understand how neural systems dynamically interact as a function of experience. In addition, we are interested in how these interactions change as a function of changing brain states such as that which occur during normal aging and pathological conditions. The neural dynamics mediating learning are also explored with computational models. The lab will be moving to the University of Washington in August, 2000. Positions can start anytime after our arrival at the UW. Interested persons should send a vita and names of 3 references to: Sheri J. Y. Mizumori Dept. Psychology 390 S. 1530 E. Rm. 502 Univ. Utah Salt Lake City, UT 84112 Office: 801-581-5555 FAX: 801-581-5841 Email: mizumori at psych.utah.edu After Aug. 1, send to: Sheri J. Y. Mizumori Dept. Psychology Box 351525 University of Washington Seattle, WA 98195-1525 From Rolf.Wuertz at neuroinformatik.ruhr-uni-bochum.de Fri Jul 7 11:33:03 2000 From: Rolf.Wuertz at neuroinformatik.ruhr-uni-bochum.de (Rolf Wuertz) Date: Fri, 7 Jul 2000 17:33:03 +0200 (MEST) Subject: JOB: Computer vision Message-ID: <200007071533.e67FX3918995@fsnif.neuroinformatik.ruhr-uni-bochum.de> The Chair of Systems Biophysics (Prof. C. von der Malsburg) at the Institute for Neurocomputing of the Ruhr-University of Bochum in Germany has vacancies for physicists, computer scientists, or electrical engineers or holders of other relevant degrees. We have a variety of exciting projects in computer vision, image understanding, and machine learning. Special emphasis is put on the study of mechanisms for learning solutions to hard vision problems from examples. The concrete projects will be tailored to match the candidates' skills and interests. The research group has been founded in 1990 and has a long record of successful projects in face recognition, gesture recognition, object recognition, robot vision, network self-organization, and models of the visual cortex. For details take a look at our WWW-site: http://www.neuroinformatik.ruhr-uni-bochum.de/ini/VDM/ Candidates should have a degree in one of the above subjects, a dedication to do active research, and an interest in computer vision. Analytical and programming skills are required, good knowledge of C++ will be an asset. Good students, who wish to pursue a diploma (Diplom) degree at our institute, are also encouraged to apply. The appointment will be for an initial period of 2 years. This can be extended, and a Ph.D. can be earned. Payment will be in accordance with BAT, the German salary scale for public employees. Please send your detailed CV including supporting material and statement of interests (email is OK) to: +----------------------------------+-----------------------------+ | Dr. Rolf P. Wuertz | Phone: +49 234 32-27994 | | Institut fuer Neuroinformatik | or: +49 234 32-27997 | | Ruhr-Universitaet Bochum | Fax: +49 234 32-14210 | | D-44780 Bochum, Germany | Room: ND03-32 | +----------------------------------+-----------------------------+ | http://www.neuroinformatik.ruhr-uni-bochum.de/ini/PEOPLE/rolf/ | | Rolf.Wuertz at neuroinformatik.ruhr-uni-bochum.de | +----------------------------------------------------------------+ From angelo at soc.plym.ac.uk Fri Jul 7 07:15:02 2000 From: angelo at soc.plym.ac.uk (Angelo Cangelosi) Date: Fri, 07 Jul 2000 12:15:02 +0100 Subject: PhD in Visualisation and Virtual Reality Message-ID: <3.0.6.32.20000707121502.007d8ca0@soc.plym.ac.uk> EPRSC PhD Studentship A PhD student is sought for an exciting and innovative project aimed at the visualisation of large data sets recorded by Neurophysiologists. The aim of this project is to develop a new approach to the analysis of experimental data in neurophysiology based on the use of computer science techniques such as graphical engineering, visualisation and virtual reality. This new approach will provide neuroscientists with an interactive environment within which to explore their data sets. The primary focus of this research is on the analysis of multi-dimensional data sets. By navigating through these large data sets, researchers will be able to focus on particular features of the data as well as identifying overall characteristics. This will support the integration of the experimental and theoretical approaches to the analysis of neurophysiological evidence. This three year studentship is funded by the UK Engineering and Physical Sciences Research Council. It is managed by Dr. L. Stuart of the Centre of Neural and Adaptive Sytsems in the School of Computing at the University of Plymouth. The student will be involved in the development of software to support the visualisation of vast quantities of neurophysiological data. Much of this development will draw on current software such as VTK and VisAD as well as the development of new representations incorporating 3D technology and/or Virtual Reality. The bursary for this studentship 6200 UK Pounds. Students are permitted to supervise/teach up to six hours per week to supplement this income. The student must have a high level of knowledge and experience of C++. Additionally, they must have good oral and written communication skills in the English language. Due to the nature of the project, the student should be able to work individually and as part of a team. For more information see http:\\www.tech.plym.ac.uk\soc\research\neural\staff\lstuart\vacancy.htm or contact Dr. L. Stuart at lstuart at plymouth.ac.uk Applications should be received before 14th April 2000, but the position will remain open until a suitable candidate is found. Candidates should send a CV, a description of motivations and the name and address of two referees to: Carole Watson School of Computing, University of Plymouth Drake Circus, Plymouth PL4 8AA, United kingdom Telephone: 01752 232541 Fax: 01752 232540 Email: c.watson at plymouth.ac.uk Thanks Liz =============================================== Dr. Liz Stuart Senior Lecturer, University of Plymouth Drake Circus, Plymouth Devon, PL4 8AA United Kingdom Telephone: +44 1752 232665 Secretary: +44 1752 232541 Fax: +44 1752 232540 =============================================== From heiko.wersing at hre-ftr.f.rd.honda.co.jp Mon Jul 10 11:41:18 2000 From: heiko.wersing at hre-ftr.f.rd.honda.co.jp (Heiko Wersing) Date: Mon, 10 Jul 2000 17:41:18 +0200 Subject: Paper and thesis available: Spatial feature binding and segmentation Message-ID: <3969EE9E.29400B1E@hre-ftr.f.rd.honda.co.jp> Dear Connectionists, I would like to announce the following recently accepted paper and my PhD Thesis on spatial feature binding and learning in competitive neural layer architectures. The paper and thesis can be downloaded from my homepage at http://www.techfak.uni-bielefeld.de/~hwersing/ Comments and questions are welcome. ----------------------------------------------------------------------- "A competitive layer model for feature binding and sensory segmentation." by Heiko Wersing, Jochen J. Steil, and Helge Ritter to appear in Neural Computation Abstract: We present a recurrent neural network for feature binding and sensory segmentation, the competitive layer model (CLM). The CLM uses topographically structured competitive and cooperative interactions in a layered network to partition a set of input features into salient groups. The dynamics is formulated within a standard additive recurrent network with linear threshold neurons. Contextual relations among features are coded by pairwise compatibilities which define an energy function to be minimized by the neural dynamics. Due to the usage of dynamical winner-take-all circuits the model gains more flexible response properties than spin models of segmentation by exploiting amplitude information in the grouping process. We prove analytic results on the convergence and stable attractors of the CLM, which generalize earlier results on winner-take-all networks, and incorporate deterministic annealing for robustness against local minima. The piecewise linear dynamics of the CLM allows a linear eigensubspace analysis which we use to analyze the dynamics of binding in conjunction with annealing. For the example of contour detection we show how the CLM can integrate figure-ground segmentation and grouping into a unified model. ------------------------------------------------------------------------ "Spatial feature binding and learning in competitive neural layer architectures" by Heiko Wersing, PhD Thesis, Faculty of Technology, University of Bielefeld, March 2000 Abstract: The goal of this thesis is to contribute to the understanding of feature binding processes by investigating an artificial neural network model for spatial feature binding, the competitive layer model (CLM). The CLM is a recurrent neural network, which employs topographically structured competitive and cooperative interactions in a system of neural layers to represent binding by the layer-wise coactivation of feature-representing neurons. This model is explored by means of mathematical analysis and simulations on artificial and challenging real-world data. The first chapter motivates the issue of feature binding as one of the important questions regarding our understanding of brain function. The second chapter gives an overview of the controversial scientific discussion of the binding problem and reviews different neural network model approaches to binding with a focus on their application in sensory segmentation and perceptual grouping. In the third chapter some new theoretical results on the stability of general linear threshold networks are established, which allow to operate the CLM in a mode of strong contextual modulation. The conditions provide a regime, where linear threshold networks are at the same time sensitive to small changes in their inputs and are capable of strong recurrent amplification without runaway activity. In chapter four the competitive layer model is introduced and its structure and dynamics are described. A deterministic annealing mechanism is introduced, which is compared to Potts-spin mean field models. The stability conditions of chapter three are applied to ensure convergence of the CLM, and additional conditions are proved, which guarantee the Winner-Take-All behaviour of the competitive columnar interactions. The grouping dynamics is characterized in relation to the lateral interactions by performing an eigensubspace analysis. Finally, the lateral coupling scheme of the CLM is generalized to general labeling problems and the chapter concludes with a comparison to other labeling approaches. Chapter five presents the application of the CLM to model a wide range of Gestalt-based perceptual grouping laws. First, the grouping of contours according to the principle of continuity is considered by using oriented edge elements as features. The results are discussed for real images, and an extension of the contour grouping approach to the task of cell segmentation is described. Other grouping principles which are considered include motion grouping, greyscale segmentation, and texture segmentation. In chapter six methods are presented to obtain appropriate lateral interactions for grouping and segmentation by supervised learning processes from manually labelled training patterns. The methods are evaluated on an artificial data set and cell images from fluorescence microscopy. A different learning approach is presented in Chapter seven, which aims at optimizing the lateral connection structure in order to improve the capabilities for solving complex constraint satisfaction problems. The proposed backtracking deterministic annealing method is interpreted as a heuristic approach to combine classical backtracking with neural deterministic annealing and successfully applied to the problem of complex tiling problems. The concluding chapter summarizes the main results and discusses possible future research directions. --------------------------------------------------------------------- -- Heiko Wersing Future Technology Research HONDA R&D EUROPE (DEUTSCHLAND) GmbH Carl-Legien-Str. 30 63073 Offenbach /Main Germany Tel.: +49-69-89011741 Fax: +49-69-89011749 e-mail: heiko.wersing at hre-ftr.f.rd.honda.co.jp From vogdrup at daimi.au.dk Mon Jul 10 03:46:25 2000 From: vogdrup at daimi.au.dk (Jakob Vogdrup Hansen) Date: Mon, 10 Jul 2000 09:46:25 +0200 Subject: PhD thesis available: Combining Predictors ... Message-ID: <200007100746.JAA10035@ppp.brics.dk> Dear Connectionists, Some people have had problems downloading my PhD thesis. I therefore give four different links to the PhD thesis in postscript and pdf format. The two last links are identical to the two first except some (unnecessary) pictures have been removed. http://www.daimi.au.dk/~vogdrup/diss.ps http://www.daimi.au.dk/~vogdrup/diss.pdf http://www.daimi.au.dk/~vogdrup/diss2.ps http://www.daimi.au.dk/~vogdrup/diss2.pdf Comments are welcome. regards, Jakob Title: Combining Predictors. Meta Machine Learning Methods and Bias/Variance & Ambiguity Decompositions Abstract: The most important theoretical tool in connection with machine learning is the bias/variance decomposition of error functions. Together with Tom Heskes, I have found the family of error functions with a natural bias/variance decomposition that has target independent variance. It is shown that no other group of error functions can be decomposed in the same way. An open problem in the machine learning community is thereby solved. The error functions are derived from the deviance measure on distributions in the one-parameter exponential family. It is therefore called the deviance error family. A bias/variance decomposition can also be viewed as an ambiguity decomposition for an ensemble method. The family of error functions with a natural bias/variance decomposition that has target independent variance can therefore be of use in connection with ensemble methods. The logarithmic opinion pool ensemble method has been developed together with Anders Krogh. It is based on the logarithmic opinion pool ambiguity decomposition using the Kullback-Leibler error function. It has been extended to the cross-validation logarithmic opinion pool ensemble method. The advantage of the cross-validation logarithmic opinion pool ensemble method is that it can use unlabeled data to estimate the generalization error, while it still uses the entire labeled example set for training. The cross-validation logarithmic opinion pool ensemble method is easily reformulated for another error function, as long as the error function has an ambiguity decomposition with target independent ambiguity. It is therefore possible to use the cross-validation ensemble method on all error functions in the deviance error family. -- Jakob V. Hansen Tlf: 86 750618 Rydevnget 87, 1. th. Kontor: B2.15 Lokal: (8942)3355 8210 Aarhus V E-mail: Vogdrup at daimi.au.dk From bengio at idiap.ch Mon Jul 10 03:26:25 2000 From: bengio at idiap.ch (Samy Bengio) Date: Mon, 10 Jul 2000 09:26:25 +0200 (MET DST) Subject: open PhD/Postdoc positions in machine learning Message-ID: Open positions for Ph.D. and Postdoctoral candidates in Machine Learning The Dalle Molle Institute for Perceptual Artificial Intelligence (IDIAP, http://www.idiap.ch) seeks qualified applicants for Ph.D. and Postdoc positions for its Machine Learning group. Many projects are currently open, such as: (a) Mixture of kernel methods such as Support Vector Machines (SVMs) and generative models such as Hidden Markov Models (HMMs) for discriminative speech processing. This work should initially build upon the preliminary works from Jaakkola and Haussler, such as, "Exploiting generative models in discriminative classifiers", in Advances in Neural Information Processing Systems 11: Proceedings of the 1998 Conference, M. Kearns, S. Solla, and D. Cohn, eds., MIT Press, 1999, pp. 487-493. (b) Several extensions of mixture and ensemble models: - Feature selection for Mixture of Experts, - Mixture of Support Vector Machines, - Mixture of binary classifiers for multiclass problems. (c) Fusion of multimodal systems at different levels: - at the score level, - for confidence intervals, - during the learning process. The ideal Ph.D. candidate should have good background in statistics, optimization, and computer science. The ideal Postdoc candidate should have strong background in statistical learning theory in general, including SVMs, neural networks, and mixture models. All applicants should be familiar with C/C++ programming under a Unix environment. Although IDIAP is located in the French part of Switzerland, English is the main working language at IDIAP. Free English and French courses are also provided. IDIAP is located in the town of Martigny (http://www.martigny.ch) in Valais, a scenic region in the south of Switzerland, surrounded by the highest mountains of Europe, and offering exciting recreational activities, including hiking, climbing and skiing, as well as varied cultural activities. It is within close proximity to Montreux (Jazz Festival) and Lausanne (EPFL, http://www.epfl.ch). Candidates should send their detailed CV to Dr. Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From Nigel.Goddard at ed.ac.uk Tue Jul 11 07:27:02 2000 From: Nigel.Goddard at ed.ac.uk (Nigel Goddard) Date: Tue, 11 Jul 2000 12:27:02 +0100 Subject: Research Programmer in Neuronal Simulation Message-ID: <396B0486.F80A2E1C@ed.ac.uk> RESEARCH PROGRAMMER IN NEURAL SIMULATION METHODS Intitute for Adaptive and Neural Computation University of Edinburgh The NEOSIM project, funded by NIMH and NSF, is developing simulation environments and associated software tools for modelling of brain processes. The project has computer science and neuroscience research goals, and aims to develop distributable, portable, parallel software for the computational neuroscience community. We seek a Research Programmer to augment the existing multidisciplinary, multinational team. Some travel to EU countries or the US will be required. The Research Programmer will be involved in design, documentation, implementation, testing, dissemination, maintenance and development of the NEOSIM framework and will also be responsible for the installation and maintenance of software packages on the research computers, and may be involved in the specification, installation and system administration of research computers including small parallel platforms. Independent research related to the goals of the NEOSIM project, including modeling, will be encouraged. Tenable for up to 3 years subject to annual renewal and further renewal beyond 3 years subject to success in attracting further funding. Interested parties should contact me as soon as possible via email to Nigel.Goddard at ed.ac.uk. I will be at CNS2000 in Brugge. Sites: http://anc.ed.ac.uk/neosim http://www.informatics.ed.ac.uk -- ============================================== Dr. Nigel Goddard Institute for Adaptive and Neural Computation Division of Informatics University of Edinburgh 5 Forrest Hill Edinburgh EH1 2QL Scotland Telephone: +44 131 650 3087 email: Nigel.Goddard at ed.ac.uk web: http://anc.ed.ac.uk/~ngoddard Fax (paper): +44 131 650 6899 eFAX (email): +1 603 698 5854 ============================================== From pelletl at CRM.UMontreal.CA Tue Jul 11 10:36:22 2000 From: pelletl at CRM.UMontreal.CA (Louis Pelletier) Date: Tue, 11 Jul 2000 10:36:22 -0400 Subject: Workshop on MEMORY, DELAYS AND MULTISTABILITY Message-ID: ************ SECOND ANNOUNCEMENT - PLEASE CIRCULATE *************** CALL FOR PAPERS As part of the activities of the 2000-2001 Theme Year in Mathematical Methods in Biology and Medicine, the Centre de Recherches Mathematiques (CRM) of the Universite de Montreal is organizing an international workshop on "MEMORY, DELAYS AND MULTISTABILITY IN NEURAL SYSTEMS" MONTREAL, 12-15 October 2000 This workshop will focus on recent advances in the mathematical modeling of neural systems having one or more of the following dynamical features: 1) memory - in particular, "dynamic" memory, 2) synaptic and propagation delays, or 3) the coexistence of steady states. These features, as well as their possible interactions, have been highlighted in a number of recent studies, e.g. of the activity of recurrent neural circuits ubiquitous in the nervous system. The emphasis will be mainly on the mathematical modeling of real biological systems, yet the workshop will also explore the relevance of these issues to disciplines ranging from psychology to artificial neural networks. MORE INFORMATION AT: http://www.crm.umontreal.ca/biomath CONTRIBUTED PAPERS BEFORE 31 AUGUST 2000: The workshop will consist of invited talks (see below) as well as short oral and poster presentations. If you wish to contribute to the workshop, your title and abstract must be sent to us before 31 August 2000 for inclusion in the programme. Your abstract should not exceed 500 words. It should be submitted electronically in either text (ASCII), Latex, or Word format to the local workshop coordinator Louis Pelletier (PELLETL at crm.umontreal.ca) with the title of the workshop in the subject heading. Please indicate your presentation format preference (oral vs poster). REGISTRATION: Registration for the conference should be done no later than 15 September 2000 for attendance and accommodation, and 30 September 2000 for attendance only. Additional information on the CRM and accommodation as well as a registration form are available at http://www.crm.umontreal.ca/biomath. Questions on the scientific aspects of the workshop should be sent to the scientific organizer at alongtin at physics.uottawa.ca (Andre Longtin). All other questions, including those concerning accommodations, should be directed to Louis Pelletier (PELLETL at crm.umontreal.ca; 514-343-2197). INVITED SPEAKERS: K. Aihara, Mathematical Engineering and Information Physics, U. Tokyo S. Becker, Psychology, McMaster U., CAN P. Bressloff, Mathematics, U. Loughborough, UK N. Brunel, Labo. de Physique Statistique, ENS, Paris S.A. Campbell, Mathematics, U. Waterloo, CAN C. Canavier, Psychology, U. New Orleans, USA G.A. Carpenter, Cognitive and Neural Systems, Boston U., USA A. Destexhe, Physiology, Laval U., CAN M. Ding, Center for Complex Systems and Brain Sciences, Florida A.U., USA W. Gerstner, Computer Science, SFIT, Lausanne L. Glass, Physiology, McGill U., CAN (*) J. Guckenheimer, Center for Applied Math., Cornell, USA A.V.M. Herz, Theoretical Biology, Humboldt, Berlin E. Izhikevich, The Neurosciences Institute, La Jolla, USA W. Maass, Tech. Univ. Graz, Austria J.G. Milton, Neurology, U. Chicago K. Pakdaman, INSERM, Paris K. Pawelzik, Institute for Theoretical Physics, U. Bremen (*)J. Rinzel, Center for Neural Science, New York U., USA M. Tsodyks, Dept. of Neurobiology, Weizmann Inst., Israel X.-J. Wang, Volen Center, Brandeis U., USA (*=tentative) From stefan.wermter at sunderland.ac.uk Thu Jul 13 03:25:03 2000 From: stefan.wermter at sunderland.ac.uk (Stefan.Wermter) Date: Thu, 13 Jul 2000 08:25:03 +0100 Subject: Intl workshop on neural networks and neuroscience Message-ID: <396D6ECE.4D067733@sunderland.ac.uk> EmerNet3: Emerging Computational Neural Network Architectures based on Neuroscience International EPSRC Workshop Date: 8-9 August 2000 Location: Durham Castle, Durham, United Kingdom Workshop web page with current information is http://www.his.sunderland.ac.uk/worksh3 Organising Committee ----------------------- Prof. Stefan Wermter Chair Hybrid Intelligent Systems Group Informatics Centre, SCET University of Sunderland Prof. Jim Austin Advanced Computer Architecture Group Department of Computer Science University of York Prof. David Willshaw Institute for Adaptive and Neural Computation Division of Informatics University of Edinburgh -------------- Contact Details --------------- Mark Elshaw (Workshop Organization) Hybrid Intelligent Systems Group Informatics Centre, SCET University of Sunderland St Peter's Way Sunderland SR6 0DD United Kingdom Phone: +44 191 515 3249 Fax: +44 191 515 2781 E-mail: Mark.Elshaw at sunderland.ac.uk Prof. Stefan Wermter (Chair) Informatics Centre, SCET University of Sunderland St Peter's Way Sunderland SR6 0DD United Kingdom Phone: +44 191 515 3279 Fax: +44 191 515 2781 E-mail: Stefan.Wermter at sunderland.ac.uk http://www.his.sunderland.ac.uk/~cs0stw/ http://www.his.sunderland.ac.uk/ From hadley at cs.sfu.ca Tue Jul 18 17:40:37 2000 From: hadley at cs.sfu.ca (Bob Hadley) Date: Tue, 18 Jul 2000 14:40:37 -0700 (PDT) Subject: Systematicity, Hebbian Learning, Lang. Acquis. In-Reply-To: <199801152116.NAA10281@css.cs.sfu.ca> from "Bob Hadley" at Jan 15, 1998 01:16:40 PM Message-ID: <200007182140.OAA10206@css.css.sfu.ca> The following paper is available at: www.cs.sfu.ca/~hadley/online.html Total pages: 37 at 1.6 spacing. Syntactic Systematicity Arising from Semantic Predictions in a Hebbian-Competitive Network BY Robert F. Hadley Adam Rotaru-Varga Dirk V. Arnold Vlad C. Cardei School of Computing Science, Simon Fraser Burnaby, B.C., V5A 1S6 Canada ABSTRACT A Hebbian-inspired, competitive network is presented which learns to predict the typical semantic features of denoting terms in simple and moderately complex sentences. In addition, the network learns to predict the appearance of syntactically key words, such as prepositions and relative pronouns. Importantly, as a by-product of the network's semantic training, a strong form of syntactic systematicity emerges. This systematicity is exhibited even at a novel, deeper level of clausal embedding. All network training is unsupervised with respect to error feedback. A novel variant of competitive learning, and an unusual hierarchical architecture are presented. The relationship of this work to issues raised by Marcus (1998) and Phillips (2000) is explored. Keywords: Systematicity, Semantic Features, Language Acquisition, Competitive Learning, Connectionism. From mzorzi at ux1.unipd.it Mon Jul 17 17:18:48 2000 From: mzorzi at ux1.unipd.it (Marco Zorzi) Date: Mon, 17 Jul 2000 23:18:48 +0200 Subject: postdoc position in connectionist modelling Message-ID: <4.3.1.0.20000717231701.00b30810@ux1.unipd.it> POST-DOCTORAL FELLOWSHIP A two-years postdoctoral position in connectionist modelling is available with Dr. Marco Zorzi and Prof. Carlo Umilte at University of Padova, Department of Psychology. The project is part of a EU-funded Research Training Network of six sites (University College London, INSERM U334 Orsay, Universite Catholique de Louvain, University of Innsbruck, University of Trieste, and University of Padova) investigating "Mathematics and the Brain". The Padova team will focus on the computational bases of mathematical cognition (basic numerical abilities and simple arithmetic). The successful applicant will have a quantitative background and excellent programming skills (MATLAB/Visual Basic/Visual C). Specific experience and/or publications in neural networks/connectionist modelling is desirable. A background in cognitive neuroscience will be a plus but is not necessary. The international network offers an excellent opportunity for gaining experience in a wide range of methodologies in cognitive neuroscience. Short visits and exchanges between laboratories are also planned. Salary and benefits are highly competitive (35000 Euro per year). Applicants must be citizens of a EU country (excluding Italy), EFTA-EEA states, Candidate States, or Israel. Application deadline is 30 September 2000, start date is negotiable (after November 1st). Send CV, reprints or preprints and the names of two references (preferably by email) to: Dr. Marco Zorzi Prof. Carlo Umilt? Dipartimento di Psicologia Generale Universit? di Padova Via Venezia 8 35131 Padova (Italy) mzorzi at ux1.unipd.it umilta at ux1.unipd.it Dr. Marco Zorzi Dipartimento di Psicologia Generale Universit? di Padova via Venezia 8 35131 Padova tel: +39 049 8276635 fax: +39 049 8276600 email: mzorzi at psico.unipd.it (and) Institute of Cognitive Neuroscience voice: +44 171 3911151 University College London fax : +44 171 8132835 17 Queen Square London WC1N 3AR (UK) http://www.psychol.ucl.ac.uk/marco.zorzi/marco.html From swatanab at pi.titech.ac.jp Wed Jul 19 01:28:32 2000 From: swatanab at pi.titech.ac.jp (Sumio Watanabe) Date: Wed, 19 Jul 2000 14:28:32 +0900 Subject: Paper : Learning theory and algebraic analysis and geometry Message-ID: <002901bff142$2e0e7c20$918a7083@deskpowerts> Dear connectionists, I would like to announce the following paper accepted for publication in Neural Computation. This paper firstly clarifies the algebraic geometrical structure of the hierarchical learning machines. Questions and comments are welcome. +++++ Title: Algebraic analysis for non-identifiable learning machines. Cite: http://watanabe-www.pi.titech.ac.jp/~swatanab/appendix.html ABSTRACT: This paper clarifies learning efficiency of a non-identifiable learning machine such as a multi-layer neural network whose true parameter set is an analytic set with singular points. By using a concept in algebraic analysis, we rigorously prove that the free energy or the Bayesian stochastic complexity is asymptotically equal to $\lambda_{1}\log n -(m_{1}-1)\log\log n + $constant, where $\lambda_{1}$ is a rational number, $m_{1}$ is a natural number, and $n$ is the number of training samples. Also we show an algorithm to calculate $\lambda_{1}$ and $m_{1}$ based on the resolution of singularities in algebraic geometry. In regular models, $2\lambda_{1}$ is equal to the number of parameters and $m_{1}=1$, whereas in non-regular models such as mutilayer networks $2\lambda_{1}$ is not larger than the number of parameters and $m_{1}\geq 1$. Since the increase of the free energy is equal to the generalization error, the non-identifiable learning machines are the better models than the regular ones if Bayesian or ensemble training is applied. +++++ Sincerely, Dr. Sumio Watanabe, Associate Professor Advanced Information Processing Division Precision and Intelligence Laboratory Tokyo Institute of Technology E-mail: swatanab at pi.titech.ac.jp (Fax)+81-45-924-5018 From ingber at ingber.com Fri Jul 21 21:30:06 2000 From: ingber at ingber.com (Lester Ingber) Date: Fri, 21 Jul 2000 20:30:06 -0500 Subject: Paper: Optimization of Trading Physics Models of Markets Message-ID: <20000721203006.A15767@ingber.com> The following preprint is available: %A L. Ingber %A R.P. Mondescu %T Optimization of Trading Physics Models of Markets %D 2001 %J IEEE Trans. Neural Networks %O Invited paper for special issue on Neural Networks in Financial Engineering. URL http://www.ingber.com/markets01_optim_trading.ps.gz ABSTRACT We describe an end-to-end real-time S&P futures trading system. Inner-shell stochastic nonlinear dynamic models are developed, and Canonical Momenta Indicators (CMI) are derived from a fitted Lagrangian used by outer-shell trading models dependent on these indicators. Recursive and adaptive optimization using Adaptive Simulated Annealing (ASA) is used for fitting parameters shared across these shells of dynamic and -- Lester Ingber http://www.ingber.com/ PO Box 06440 Wacker Dr PO Sears Tower Chicago IL 60606-0440 http://www.alumni.caltech.edu/~ingber/ From terry at salk.edu Fri Jul 21 19:50:19 2000 From: terry at salk.edu (Terry Sejnowski) Date: Fri, 21 Jul 2000 16:50:19 -0700 (PDT) Subject: NEURAL COMPUTATION 12:8 Message-ID: <200007212350.QAA03312@dax.salk.edu> Neural Computation - Contents - Volume 12, Number 8 - August 1, 2000 ARTICLE Neural Systems as Nonlinear Filters Wolfgang Maass and Eduardo D. Sontag NOTE Analytical Model for the Effects of Learning on Spike Count Distributions Gianni Settanni and Alessandro Treves LETTERS Calculation of Interspike Intervals for Integrate-and-Fire Neurons with Poisson Distribution of Synaptic Inputs A. N. Burkitt and G. M. Clark Statistical Procedures for Spatio-Temporal Neuronal Data with Applications to Optical Recording of the Auditory Cortex O. Francois, L. Mohamed Abdallahi, J. Horikawa, I Taniguchi and T. Herve Probabilistic Motion Estimation Based on Temporal Coherence Pierre-Yves Burgi, Alan L. Yuille, and Norberto M. Grzywacz Boosting Neural Networks Holger Schwenk and Yoshua Bengio Gradient-based Optimization of Hyper-parameters Yoshua Bengio A Signal-Flow-Graph Approach to On-line Gradient Calculation Paolo Campolucci, Aurelio Uncini and Francesco Piazza Bootstrapping Neural Networks Jurgen Franke and Michael H. Neumann Information Geometry of Mean-Field Approximation Toshiyuki Tanaka Measuring the VC-Dimension Using Optimized Experimental Design Xuhui Shao, Vladimir Cherkassky and William Li ----- ON-LINE - http://neco.mitpress.org/ SUBSCRIPTIONS - 2000 - VOLUME 12 - 12 ISSUES USA Canada* Other Countries Student/Retired $60 $64.20 $108 Individual $88 $94.16 $136 Institution $430 $460.10 $478 * includes 7% GST MIT Press Journals, 5 Cambridge Center, Cambridge, MA 02142-9902. Tel: (617) 253-2889 FAX: (617) 258-6779 mitpress-orders at mit.edu ----- From bengio at idiap.ch Wed Jul 26 09:33:56 2000 From: bengio at idiap.ch (Samy Bengio) Date: Wed, 26 Jul 2000 15:33:56 +0200 (MET DST) Subject: Open positions in speech, computer vision, machine learning, and multimodal interfaces Message-ID: SEVERAL OPEN POSITIONS IN SPEECH, COMPUTER VISION, MACHINE LEARNING, AND MULTIMODAL INTERFACES In view of a massive growth of the Institute and the development of new activites oriented towards advanced multimodal interfaces, the Dalle Molle Institute for Perceptual Artificial Intelligence (IDIAP, http://www.idiap.ch), IDIAP currently welcomes applications of talented candidates at all levels with expertise or strong interest in the fields of speech processing, computer vision, machine learning, and multimodal interaction. The open positions include: management and senior positions (including one scientific deputy director and one speech processing group leader), project leaders, postdocs, and PhD students. It is expected that a few tenure track positions at the Assistant Professor level could also become available. Preference will be given to candidates with experience in one or several of the following areas: signal processing, statistical pattern recognition (typically applied to speech and scene analysis), neural networks, hidden Markov models, speech and speaker recognition, computer vision, human/computer interaction (dialog). Senior and postdoc candidates should also have a proven record of high quality research and publications. All applicants should be experienced in C/C++ programming and familiar with the Unix environment; they should also be able to speak and write in English (and be willing to learn French). ABOUT IDIAP IDIAP is a semi-private research institute affiliated with the Swiss Federal Institute of Technology at Lausanne (EPFL) and the University of Geneva. Located in Martigny (Valais, CH), IDIAP is partly funded by the Swiss Federal Government, the State of Valais, and the City of Martigny, and is involved in numerous national and international (European) projects. IDIAP is mainly carrying research and development in the fields of speech and speaker recognition, computer vision, and machine learning, and is recognized as a university level research laboratory, involving permanent senior scientists, postdocs, and PhD students (usually awarded an EPFL degree). IDIAP currently numbers around 35-40 scientists. LOCATION IDIAP is located in the town of Martigny (http://www.martigny.ch) in Valais, a scenic region in the South of Switzerland, surrounded by the highest mountains of Europe, and offering exciting recreational activities (including hiking, climbing and skiing), as well as varied cultural activities. It is also within close proximity to Montreux, Lausanne (EPFL) and Lake Geneva, and centrally located for travel to other parts of Europe. PROSPECTIVE CANDIDATES should send their detailed CV to: Prof. Herve Bourlard Director of IDIAP P.O. Box 592, Simplon, 4 CH-1920 Martigny, Switzerland Email: bourlard at idiap.ch Phone: +41-27-721.77.20; Fax: +41-27-721.77.12 ----- Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From bengio at idiap.ch Tue Jul 25 08:49:37 2000 From: bengio at idiap.ch (Samy Bengio) Date: Tue, 25 Jul 2000 14:49:37 +0200 (MET DST) Subject: SVMTorch: A new SVM program for Large-Scale Regression and Classification Problems Message-ID: I would like to inform you of the following new SVM software for large-scale regression and classification problems, available at http://www.idiap.ch/learning/SVMTorch.html. Information about this new software follows: SVMTorch A Support Vector Machine for Large-Scale Regression and Classification Problems Ronan Collobert (collober at idiap.ch) IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland Description SVMTorch is a new implementation of Vapnik's Support Vector Machine that works both for classification and regression problems, and that has been specifically tailored for large-scale problems (such as more than 20000 examples, even for input dimensions higher than 100). Source Code The source code is free for academic use. It must not be modified or distributed without prior permission the author. When using SVMTorch in your scientific work, please cite the following article: Ronan Collobert and Samy Bengio, Support Vector Machines for Large-Scale Regression Problems, IDIAP-RR-00-17, 2000. (available at ftp://ftp.idiap.ch/pub/reports/2000/rr00-17.ps.gz). The software has been successfully compiled on Sun/SOLARIS, Intel/LINUX and Alpha/OSF operating systems. Your can download it from ftp://ftp.idiap.ch/pub/learning/SVMTorch.tgz. Try It !!! First, you should download the source code from ftp://ftp.idiap.ch/pub/learning/SVMTorch.tgz and the examples from ftp://ftp.idiap.ch/pub/learning/TrainData.tgz. Put this two archive files in the same directory, and decompress them with zcat SVMTorch.tgz | tar xf - zcat TrainData.tgz | tar xf - It creates two new directories : "SVMTorch" and "TrainData". Now, go in the "SVMTorch" directory and edit the Makefile. You should only have to change the following lines, depending on your specific platform : # C-compiler #CC=gcc CC=cc # C-Compiler flags #CFLAGS=-Wall -W -O9 -funroll-all-loops -finline -fomit-frame-pointer -ffast-math CFLAGS=-native -fast -xO5 # linker #LD=gcc LD=cc # linker flags #LFLAGS=-Wall -W -O9 -funroll-all-loops -finline -fomit-frame-pointer -ffast-math LFLAGS=-native -fast -xO5 # libraries LIBS=-lm The default configuration is set for a machine running with the Sun Workshop compiler. An alternate (commented) configuration is proposed for the GNU gcc compiler. Type "make all" and pray. It should compile without any warning. For some platform, you could have to change the include files needed for "times", a non-standard function used by svm_torch. You would have to edit the file "general.h" and change the lines #ifdef I_WANT_TIME #include /*#include */ #include #endif If it doesn't work or if you don't want to measure the time of the learning machine, just comment the line : #define I_WANT_TIME Note that in "general.h" you can comment the line #define USEDOUBLE in order to do the computations in float. IT'S A BAD IDEA : svm_torch needs precision. If everything went well, you should have two programs : "svm_torch" and "svm_test". The first one is the learning machine and the second one is the testing machine. If you want to show all the options, just run svm_torch or svm_test without any parameter. To test the program in classification, try : svm_torch -v -ae ../TrainData/classif_train.dat ../TrainData/model_dummy It takes less than two minutes on a 300Mhz computer. You should have around 914 support vectors (this number could slightly change depending on the precision of your machine). To test the SVM on the train data, try : svm_test -ae ../TrainData/model_dummy ../TrainData/classif_train.dat You should have around 0.78% missclassified. To test the program in regression, try : svm_torch -v -ae -rm -st 900 -eps 20 ../TrainData/regress_train.dat ./TrainData/model_dummy You should have around 597 support vectors. Test the model with : svm_test -ae ../TrainData/model_dummy ../TrainData/regress_train.dat The mean squared error should be around 187.2. Options The general syntax of svm_torch and svm_test is svm_torch [options] example_file model_file svm_test [options] model_file test_file Where "example_file" is your training set file, "test_file" is your testing set file and "model_file" is the SVM-model created by svm_torch. All options are described when you launch svm_torch or svm_test without any argument. By default, svm_torch is a classification machine. If you want the regression machine, use option -rm. You should always use option -v with svm_torch : it gives a current error during learning. This error is only an indicator. It can oscillate. File format There are two main input formats for "input_file" and "test_file" in SVMTorch : an ASCII format, and a binary one. The ASCII format is the following: .... . . . .... where is an ASCII floating point number corresponding to the j-th value of the i-th example and is the i-th desired output (in classification, it should be +1/-1). With the same notation, the binary format is: ... ....... ... ... (First save the input table, then the output table, all in binary) There is another special input format for svm_test, when you don't have the desired output. (To use with the -no option). The ASCII version of this format is : .... . . . .... And the binary version is : ... ....... ... ----- Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From giro at cortex.hut.fi Tue Jul 25 04:43:00 2000 From: giro at cortex.hut.fi (Mark Girolami) Date: Tue, 25 Jul 2000 11:43:00 +0300 (EEST) Subject: PhD Studentship Available Message-ID: PhD Studentship Available Applied Computational Intelligence Research Unit University of Paisley Scotland, UK. & Power Technology Research Centre POWERGEN (UK) Nottingham, UK. Title of Investigation: 'The Discrimination of Novel Complex-Normal Behaviour from Faulty Operating Conditions for Large Rotodynamic Machines' A three year fully funded Phd studentship is available in the Applied Computational Intelligence Research Unit at the University of Paisley. This is a joint research project with POWERGEN (UK) and it is envisaged that the student will work over certain periods of time at the Power Technology Centre in Nottingham. Project Brief: The automated discrimination of genuine fault conditions from acceptable operating behaviour of large and complex rotodynamic machines is a challenging engineering problem. This is effectively a classification task, however as the available training samples consist largely of acceptable operating conditions and the occurrence of genuine fault conditions is relatively rare we have in effect to consider the problem of novelty detection. Standard classification techniques are not suitable for this particular problem as faults will be potentially diverse and ill sampled. There are also multi-state situations when a machine can be in a number of distinct states depending on operating history. There have been attempts in the past to use regression and neural network methods to try to model the relationship between machine vibration and operating conditions, but with limited success. The development and application of support vector/kernel methods to this problem domain is proposed as a potential avenue of investigation. Applications from prospective candidates with backgrounds in either mathematics, statistics, physics, engineering are especially welcome. Informal Enquiries can be made to: Prof. M.A. Girolami Chair of Computational Intelligence Department of Computing and Information Systems University of Paisley High Street Paisley, PA1 2BE Scotland, UK. e-mail: mark.girolami at paisley.ac.uk From giro at james.hut.fi Tue Jul 25 04:35:46 2000 From: giro at james.hut.fi (Mark Girolami) Date: Tue, 25 Jul 2000 11:35:46 +0300 (EEST) Subject: Advances in Independent Component Analysis Message-ID: New Book Now Available: Title: Advances in Independent Component Analysis Editor : Mark Girolami Publisher : Springer Verlag ISBN 1-85233-263-8 Companion Website : http://www.dcs.ex.ac.uk/ica Since 1995 the neural computing research community has produced a large number of publications dedicated to the Blind Separation of Sources (BSS) or Independent Component Analysis (ICA). This book is the outcome of a workshop which was held after the 1999 International Conference on Artificial Neural Networks (ICANN) in Edinburgh, Scotland. Some of the most active and productive neural computing researchers gathered to present and share new ideas, models and experimental results and this volume documents the individual presentations. Individual Chapter Titles 1. Hidden Markov Independent Component Analysis Authors : W. Penny, R.Everson, S. Roberts 2. Particle Filters for Non-Stationary ICA Authors : R. Everson, S.Roberts 3. Analysing the Independence of the Components by Topography Authors : A.Hyvarinen, P. Hoyer, M. Inki. 4. Dependent Component Analysis Author : A Barros 5. Ensemble Learning Authors : H. Lappalainen, J. Misken 6. Bayesian Non-linear ICA by MLP's Authors: H. Lappalainen, A. Honkela. 7. Ensemble Learning for Blind Separation and Deconvolution Authors : J.Misken, D. MacKay. 8. MUCICA for Rank-Deficient Distributions Authors : F. Palmieri, A.Budillon. 9. Blind Separation of Noisy image Mixtures Author : L.K. Hansen. 10. Searching for independence in Electromagnetic Brain Waves Authors : R.Vigario, J. Sarela, E. Oja. 11. ICA on Noisy Data : A Factor Analysis Approach Author : S Ikeda. 12. Analysis of Optical Imaging Data Using Weak Models and ICA Authors :J. Porril, J. Stone, J. Berwick, J. Mayhew, P. Coffey. 13. Independent Components in Text. Authors : T. Kolenda, L. Hansen, S.Sigurdsson. 14. Seeking Independence Using Biologically-Inspired ANN's Authors : P.Lai, D. Charles, C. Fyfe. From ted.carnevale at yale.edu Sun Jul 23 17:28:18 2000 From: ted.carnevale at yale.edu (Ted Carnevale) Date: Sun, 23 Jul 2000 17:28:18 -0400 Subject: NEURON course at SFN 2000 meeting Message-ID: <397B6372.515185E1@yale.edu> Short Course Announcement USING THE NEURON SIMULATION ENVIRONMENT Satellite Symposium, Society for Neuroscience Meeting 9 AM - 5 PM on Saturday, Nov. 4, 2000 Speakers: N.T. Carnevale, M.L. Hines, J.W. Moore, and G.M. Shepherd This 1 day course with lectures and live demonstrations will present information essential for teaching and research applications of NEURON, an advanced simulation environment that handles realistic models of biophysical mechanisms, individual neurons, and networks of cells. The emphasis is on practical issues that are key to the most productive use of this powerful and convenient modeling tool. Features that will be covered include: importing detailed morphometric data constructing and managing models of neurons with the CellBuilder constructing and managing network models with the new Network Builder using the Multiple Run Fitter to optimize models that have high-dimensional datasets NEURON's new functions for parallelizing simulations over a workstation cluster (32-bit MSWindows and/or UNIX/Linux boxes) database resources for empirically-based modeling Each registrant will receive a CD-ROM with software, plus a comprehensive set of notes that includes material which has not appeared elsewhere in print. For more information see the course's WWW pages at http://www.neuron.yale.edu/easy2k.html --Ted Supported in part by the National Science Foundation. Opinions expressed are those of the authors and not necessarily those of the Foundation. From pelillo at dsi.unive.it Fri Jul 21 03:28:28 2000 From: pelillo at dsi.unive.it (Marcello Pelillo) Date: Fri, 21 Jul 2000 09:28:28 +0200 (MET DST) Subject: An IEEE-TNN paper on Max-Weight-Clique In-Reply-To: Message-ID: Dear all, the following paper, accepted for publication in the IEEE Transactions on Neural Networks, is accessible at the following www site: http://www.dsi.unive.it/~pelillo/papers/ieee-tnn-mwcp.ps.gz Comments and suggestions are welcome! Best regards, Marcello Pelillo ********************* Approximating the Maximum Weight Clique Using Replicator Dynamics I. M. Bomze (University of Vienna, Austria) M. Pelillo (University of Venice, Italy) V. Stix (University of Vienna, Austria) Abstract Given an undirected graph with weights on the vertices, the maximum weight clique problem (MWCP) is to find a subset of mutually adjacent vertices (i.e., a clique) having largest total weight. This is a generalization of the classical problem of finding the maximum cardinality clique of an unweighted graph, which arises as a special case of the MWCP when all the weights associated to the vertices are equal. The problem is known to be NP-hard for arbitrary graphs and, according to recent theoretical results, so is the problem of approximating it within a constant factor. Although there has recently been much interest around neural network algorithms for the unweighted maximum clique problem, no effort has been directed so far towards its weighted counterpart. In this paper, we present a parallel, distributed heuristic for approximating the MWCP based on dynamics principles developed and studied in various branches of mathematical biology. The proposed framework centers around a recently introduced continuous characterization of the MWCP which generalizes an earlier remarkable result by Motzkin and Straus. This allows us to formulate the MWCP (a purely combinatorial problem) in terms of a continuous quadratic programming problem. One drawback associated with this formulation, however, is the presence of "spurious" solutions, and we present characterizations of these solutions. To avoid them we introduce a new regularized continuous formulation of the MWCP inspired by previous works on the unweighted problem, and show how this approach completely solves the problem. The continuous formulation of the MWCP naturally maps onto a parallel, distributed computational network whose dynamical behavior is governed by the so-called replicator equations. These are dynamical systems introduced in evolutionary game theory and population genetics to model evolutionary processes on a macroscopic scale. We present theoretical results which guarantee that the solutions provided by our clique finding replicator network are actually the ones being sought. Extensive experiments on both randomly generated and standard benchmark graphs have been conducted, and the results obtained confirm the effectiveness of the proposed approach. Key words: Maximum weight clique, replicator equations, evolutionary game theory, dynamical systems, quadratic programming. ********************* ________________________________________________________________________ Marcello Pelillo Dipartimento di Informatica Universita' Ca' Foscari di Venezia Via Torino 155, 30172 Venezia Mestre, Italy Tel: (39) 041 2908.440 Fax: (39) 041 2908.419 E-mail: pelillo at dsi.unive.it URL: http://www.dsi.unive.it/~pelillo From movellan at markov.ucsd.edu Tue Jul 18 20:52:54 2000 From: movellan at markov.ucsd.edu (movellan) Date: Tue, 18 Jul 2000 17:52:54 -0700 Subject: Tech Reports Message-ID: <3974FBE6.AC730EFD@markov.ucsd.edu> The following technical reports are available online at http://mplab.ucsd.edu (follow links to Tech Reports) A Multithreaded Approach to Real Time Face Tracking Boris E. Shpungin and Javier R. Movellan We present a multithreaded approach to real-time face tracking. The system consists of a set of modules running on different processing threads and coordinated by a mini operating system. The first module uses color, shape, and face dynamics to determine a 2-D region of interest. The second module is a mixture of linear experts classifier whose task is to help determine whether the region of interest is indeed a face. ======================= Active Inference in Concept Induction Jonathan D. Nelson and Javier R. Movellan People are active information gatherers, constantly experimenting and seeking information relevant to their goals. A reasonable approach to active information gathering is to ask questions and conduct experiments that maximize the expected information gain given our current beliefs (Lindley, 1956; Good, 1966; MacKay, 1992). In this paper we compare the behavior of human subjects with that of an optimal information-gathering agent (infomax) in a concept induction task (Tennenbaum, 1999). Results show high consistency between subjects in their information gathering behavior. However infomax generally fails to predict what subjects do. It is unclear at this point whether the failure of infomax is due to problems with Tennenbaum's concept induction model, or due to the fact that subjects use suboptimal heuristics (e.g., confirmatory sampling). From C.Campbell at bristol.ac.uk Tue Jul 18 14:12:42 2000 From: C.Campbell at bristol.ac.uk (Colin Campbell) Date: Tue, 18 Jul 2000 19:12:42 +0100 (BST) Subject: Postdoctoral Position/Computational Intelligence Message-ID: EPSRC Postdoctoral Position available in Computational Intelligence Support Vector Machines and other Kernel-based Learning Methods Applications are invited for a EPSRC funded 2 year postdoctoral position in computational intelligence. The principal area of investigation will be kernel methods (for example, Support Vector Machines) and their applications. The emphasis of the proposed research will be the development of new algorithms (e.g. for novelty detection, classification, query learning, etc) and theoretical work to validate this approach. The project will also include some investigation of the application of kernel methods to bioinformatics, medical decision support and engineering in collaboration with research groups in these areas. The position is open to citizens of any nationality. Further details about the proposed research area may be obtained from our webpage http://lara.enm.bris.ac.uk/cig which has a downloadable review paper (An Introduction to Kernel Methods) describing the subject in more detail. The closing date for applications is *** 20th August 2000 *** Further details can be obtained from: Dr. Colin Campbell, Dept. of Engineering Mathematics, Queen's Building, University of Bristol, Bristol BS8 1TR, United Kingdom Email: C.Campbell at bris.ac.uk Tel: +44 (0) 117 928 9858 Fax: +44 (0) 117 925 1154 From mkm at hnc.com Thu Jul 20 19:48:24 2000 From: mkm at hnc.com (McClarin, Melissa) Date: Thu, 20 Jul 2000 16:48:24 -0700 Subject: HNC Financial Solutions seeks a Staff Scientist Message-ID: <72A838A51366D211B3B30008C7F4D36303FEEBDE@pchnc.hnc.com> At HNC Software, innovation and growth make us one of the most dynamic software companies in the world. We have been recognized by Forbes (ASAP Dynamic 100), Fortune (100 Fastest growing companies in America), and Software Magazine (500 Top Software Companies). Also, the San Diego Business Journal listed us as one of "The Best Companies to Work for in San Diego." HNC Financial Solutions, a division of HNC Software, is a world leader in the development and delivery of predictive, neural networks based software solutions for the financial industry. We offer the opportunity to work with cutting edge technology and a world class team in a casual atmosphere. Benefits include three weeks vacation, stock option grants and tuition reimbursement. HNC Financial Solutions is seeking a full time Staff Scientist to be based at our headquarters in San Diego, California. Please see job description and requirements below. To apply, please use one of the following methods: Mail: 5935 Cornerstone Ct. W San Diego, CA 92121 Email: hireme at hnc.com Online:http://www.hnc.com To apply for this position, please reference job code F142SD Staff Scientist Duties/Job Description: Responsibilities include designing and building predictive models based on the latest technologies in neural networks, pattern recognition/artificial intelligence, and statistical modeling for various applications in the financial industry. Specific responsibilities may vary by projects but will include analyzing data to determine suitability for modeling, pattern identifications and feature (variable) selection from large amounts of data, experimenting with different types of models, analyzing performance, and reporting results to customers. Required Qualifications (Experience/Skills): MS or PHD in Computer Science, Electrical Engineering, Applied Statistics/Math, or related field. Minimum 2 years of experience in pattern recognition, mathematical modeling, or data analysis on real world problems. Familiarity with the latest modeling techniques and tools. Good oral and written communication skills, both in terms of interacting with customers and co-workers. Very comfortable with UNIX and C. Familiar with SAS, or other analysis tools. Preferred Qualifications (Experience/Skills): Strong mathematical appetite, problem solving and computer skills. Good UNIX scripting and rapid prototyping skill. Quick learner and a good team player. Experience in designing systems based on neural networks, pattern recognition, and/or statistical modeling techniques for financial, health care, marketing, or other real world applications. From danny.silver at acadiau.ca Thu Jul 27 12:35:06 2000 From: danny.silver at acadiau.ca (Daniel L. Silver) Date: Thu, 27 Jul 2000 13:35:06 -0300 Subject: "Daniel L. Silver": PhD thesis available: Selective Transfer of Neural Network Task Knowledge Message-ID: <200007271637.e6RGbUK29773@garlic.acadiau.ca> Dear Connectionists, This is to announce the availability of my PhD thesis for download. Title: Selective Transfer of Neural Network Task Knowledge Postscript URL: http://dragon.acadiau.ca/~dsilver/DLS_THESIS.PS Zip of PS URL: http://dragon.acadiau.ca/~dsilver/DLS_THESIS.zip (323 pages) Best regards, Danny Silver ============== Keywords: task knowledge transfer, artificial neural networks, sequential learning, inductive bias, task relatedness, knowledge based inductive learning, learning to learn, knowledge consolidation Abstract: Within the context of artificial neural networks (ANN), we explore the question: How can a learning system retain and use previously learned knowledge to facilitate future learning? The research objectives are to develop a theoretical model and test a prototype system which sequentially retains ANN task knowledge and selectively uses that knowledge to bias the learning of a new task in an efficient and effective manner. A theory of {\em selective functional transfer} is presented that requires a learning algorithm that employs a {\em measure of task relatedness}. $\eta$MTL is introduced as a knowledge based inductive learning method that learns one or more secondary tasks within a back-propagation ANN as a source of inductive bias for a primary task. $\eta$MTL employs a separate learning rate, $\eta_k$, for each secondary task output $k$. $\eta_k$ varies as a function of a measure of relatedness, $R_k$, between the $k^{th}$ secondary task and the primary task of interest. Three categories of {\em a priori} measures of relatedness are developed for controlling inductive bias. The {\em task rehearsal method} (TRM) is introduced to address the issue of sequential retention and generation of learned task knowledge. The representations of successfully learned tasks are stored within a {\em domain knowledge} repository. {\em Virtual training examples} generated from domain knowledge are rehearsed as secondary tasks in parallel with each new task using either standard multiple task learning (MTL) or $\eta$MTL. TRM using $\eta$MTL is tested as a method of selective knowledge transfer and sequential learning on two synthetic domains and one medical diagnostic domain. Experiments show that the TRM provides an excellent method of retaining and generating accurate functional task knowledge. Hypotheses generated are compared statistically to single task learning and MTL hypotheses. We conclude that selective knowledge transfer with $\eta$MTL develops more effective hypotheses but not necessarily with greater efficiency. The {\em a priori} measures of relatedness demonstrate significant value on certain domains of tasks but have difficulty scaling to large numbers of tasks. Several issues identified during the research indicate the importance of consolidating a representational form of domain knowledge. ====================================================== Daniel L. Silver Danny.Silver at AcadiaU.ca Assistant Professor Intelligent Information Technology Research Centre Jodery School of Computer Science, Office 315 Acadia University ph:(902)585-1105 fax:(902)585-1067 Wolfville, NS B0P 1X0 From oby at cs.tu-berlin.de Fri Jul 28 09:18:52 2000 From: oby at cs.tu-berlin.de (Klaus Obermayer) Date: Fri, 28 Jul 2000 15:18:52 +0200 (MET DST) Subject: positions available Message-ID: <200007281318.PAA01731@pollux.cs.tu-berlin.de> Postdoctoral and Graduate Student Positions Available Neural Information Processing Group, Department of Computer Science Technical University of Berlin, Berlin, Germany The Neural Information Processing Group anticipates several openings starting this fall: 1 postdoctoral position, salary level BAT IIa 1 graduate student position, salary level, 2/3 BAT IIa Both positions are available for candidates, who are interested in developing algorithms and tools for: - 3D image processing of confocal microscope data - processing of functional imaging data (Ca-imaging, optical recording, fMRI) The candidates are expected to join a BMBF-sponsored collaboration with two computer science and four experimental neurobiology groups. We expect the postdoctoral position to be available as early as Sept. 1st 2000. The position will be initially for one year, but an extension up to three years is possible. The graduate student position is available immediately for a duration of initially two years. 2 graduate student positions, salary level BAT IIa: The positions are for candidates, who are interested in the following research areas: - computational models of visual cortex - theory of unsupervised learning algorithms Teaching duties include 4 hours tutoring per week (in German) during the winter and summer terms, i.e.\ the successful candidate must be fluent in the German language. Both positions are available Oct. 1st 2000 for a duration of five years maximum. The CS department of the Technical University of Berlin approves students with foreign Masters or Diplom degrees for graduate study when certain standards w.r.t. the subject, grades, and the awarding institution are met. The university also allows for a thesis defense in the English language. Interested candidates please send their CV, transcripts of their certificates, a short statement of their research interest, and a list of publications to: Prof. Klaus Obermayer FR2-1, NI, Informatik, Technische Universitaet Berlin Franklinstrasse 28/29, 10587 Berlin, Germany phone: ++49-30-314-73120, fax: -73121, email: oby at cs.tu-berlin.de prefereably by email. For a list of relevant publications and an overview of current research projects please refer to our web-page at: http://ni.cs.tu-berlin.de/ Cheers Klaus --------------------------------------------------------------------------- Prof. Dr. Klaus Obermayer phone: 49-30-314-73442 FR2-1, NI, Informatik 49-30-314-73120 Technische Universitaet Berlin fax: 49-30-314-73121 Franklinstrasse 28/29 e-mail: oby at cs.tu-berlin.de 10587 Berlin, Germany http://ni.cs.tu-berlin.de/ From gary at cs.ucsd.edu Fri Jul 28 13:23:30 2000 From: gary at cs.ucsd.edu (Gary Cottrell) Date: Fri, 28 Jul 2000 10:23:30 -0700 (PDT) Subject: Post doctoral positions in the PEN network Message-ID: <200007281723.KAA18891@gremlin.ucsd.edu> Postdoctoral Research Positions in Cognitive Neuroscience Applications are invited for four 2-year postdoctoral positions starting between 1/1/1 and 1/1/2 in the Perceptual Expertise Network (PEN), recently funded by the James S. McDonnell Foundation. There is no application deadline, and appropriate applicants will be chosen as they apply. The network's goals are to explore how "different" brains approach object recognition and categorization by comparing normals, agnosics, autistic individuals, non-human primates, and computational models. The network is headed by Isabel Gauthier (Vanderbilt University) and includes the following PIs: James Tanaka (Oberlin College), Tim Curran (Case Western University), Michael Tarr (Brown University), Marlene Behrmann (Carnegie Mellon University), Daniel Bub (University of Victoria), Robert Schultz (Yale Child Study Center), David Sheinberg (Brown University) and Garrison Cottrell (UCSD). One of the positions is to work at Vanderbilt with Dr. Gauthier, using fMRI and psychophysics to study the neural bases of visual object recognition. Two other positions may be held at any of the network's sites, with preference given to applicants wishing to spend time at multiple sites and demonstrating enthusiasm and skill for integrative and collaborative work. An additional 2-year position is available at the University of Victoria (under Dr. Bub's supervision), beginning 1/1/2. Applicants should have strong interests in detailed analyses of neuropsychological cases. All applicants should demonstrate an outstanding background in conducting original research on the mechanisms of face and/or object recognition, or work that has theoretical relevance in this field. For all positions, the network will support attendance to triannual PEN meetings. Post-doctoral PEN fellows will have access to a many cutting-edge techniques in cognitive neuroscience, including a computer graphics designer for stimuli preparation, fMRI, ERP, computational modeling, psychophysics, neurophysiology and neuropsychological populations (individuals with visual agnosia or autism spectrum disorders). Applicants for these positions should submit C.V., relevant reprints, 3 letters of references and a 1-2 page proposal on how they hope to combine the techniques or approaches of at least two of the PEN investigators to pursue issues in visual object representation and expertise. Applications in electronic format are encouraged (Word98 or PDF) and letters of recommendation can be sent electronically to isabel.gauthier at vanderbilt.edu. Applications can also be sent by mail to: Isabel Gauthier, Ph.D. Assistant Professor, Psychology Departmen, 502 Wilson Hall, Vanderbilt University, Nashville, TN 37240. Phone: 615 322 1778 FAX: 615 343 8449, isabel.gauthier at vanderbilt.edu. More information can be found at: http://www.psy.vanderbilt.edu/faculty/gauthier/PENpostdocs.html cheers, gary Gary Cottrell 858-534-6640 FAX: 858-534-7029 Faculty Assistant Karen Chang: 858-822-3286 kkchang at cs.ucsd.edu Computer Science and Engineering 0114 IF USING FED EX INCLUDE THE FOLLOWING LINE: "Only connect" 3101 Applied Physics and Math Building University of California San Diego -E.M. Forster La Jolla, Ca. 92093-0114 Email: gary at cs.ucsd.edu or gcottrell at ucsd.edu Home page: http://www-cse.ucsd.edu/~gary/ From: esann To: "Connectionists at cs.cmu.edu" References: From bogus@does.not.exist.com Fri Jul 28 14:30:23 2000 From: bogus@does.not.exist.com () Date: Fri, 28 Jul 2000 20:30:23 +0200 Subject: Call for sessions ESANN'2001 European Symposium on Artificial Neural Networks Message-ID: ---------------------------------------------------- | | | ESANN'2001 | | | | 8th European Symposium | | on Artificial Neural Networks | | | | Bruges (Belgium) - April 25-26-27, 2001 | | | | Call for special sessions | ---------------------------------------------------- The organizing committee of the ESANN'2001 conference is looking for proposals or suggestions concerning the organization of special sessions during the conference. Details on the organization of special sessions are available at the URL http://www.dice.ucl.ac.be/esann/callforspecialsessions.htm. More information about the ESANN'2001 conference is available at http://www.dice.ucl.ac.be/esann/. The ESANN'2001 conference will focus on fundamental aspects of ANNs: theory, models, learning algorithms, mathematical aspects, approximation of functions, classification, control, time-series prediction, statistics, signal processing, vision, self-organization, vector quantization, evolutive learning, psychological computations, biological plausibility, etc. Papers on links and comparisons between ANNs and other domains of research (such as statistics, data analysis, signal processing, biology, psychology, evolutive learning, bio-inspired systems, etc.) are encouraged. The conference will be held in Bruges (also called "Venice of the North"), one of the most beautiful medieval towns in Europe, with easy access to international travel connections. For any proposal and suggestion, please send an e-mail before August 15, 2000 to mailto:esann at dice.ucl.ac.be. ===================================================== ESANN - European Symposium on Artificial Neural Networks http://www.dice.ucl.ac.be/esann * For submissions of papers, reviews,... Michel Verleysen Univ. Cath. de Louvain - Microelectronics Laboratory 3, pl. du Levant - B-1348 Louvain-la-Neuve - Belgium tel: +32 10 47 25 51 - fax: + 32 10 47 25 98 mailto:esann at dice.ucl.ac.be * Conference secretariat D facto conference services 27 rue du Laekenveld - B-1080 Brussels - Belgium tel: + 32 2 420 37 57 - fax: + 32 2 420 02 55 mailto:esann at dice.ucl.ac.be ===================================================== From rjb at biols.susx.ac.uk Mon Jul 31 12:20:16 2000 From: rjb at biols.susx.ac.uk (Roland Baddeley) Date: Mon, 31 Jul 2000 17:20:16 +0100 Subject: Statistical analysis of visual signaling in Cuttlefish Message-ID: <3985A740.24D278EA@biols.susx.ac.uk> Three year postdoctoral position funded by the BBSRC to work on the computational analysis of visual signalling in cuttlefish, to work in collaboration with Dr Daniel Osorio and Dr Roland Baddeley at Sussex University. Cuttlefish have a remarkable ability to change their skin coloration patterns giving them amazing camouflage abilities, and a very effective visual communication channel. Previous work on cuttlefish has relied on qualitative classification of their body patterns, but by using digital video cameras, image warping techniques, and statistical pattern analysis, it is now possible to quantitatively analyze these changing skin patterns. Doing this we hope to understand their role in camouflage and communication. We are therefore looking for a postdoctoral researcher (for up to 3 years) who will: 1. Analyze the skin patterns generated by the cuttlefish. We have developed warping and independent components analysis software. The researcher would be expected implement additional means of analysis that would shed light on the dimensionality, any clustering of the patterns (expected), and implement methods to study the dynamics of the change. 2. Conduct experiments on camouflage. The animals are kept in tanks where ``textures'' can be inserted underneath. We can therefore analyze camouflage by quantifying how the cuttlefish maps the texture pattern inserted, onto its skin patterns. This will give unprecedented insight into texture and form perception by a non-human animal. As well as conducting the experiments, the researcher would be expected to maintain the animals. No previous experience is required as training will be provided. 3. Conduct experiments on communication. The cuttlefish communicate by generating stereotypical patterns on the skin. As well as analyzing these patterns on their own, given a quantitative description, the relationship between the patterns of two communicating animals can be analyzed (information capacity, relationship to behavior). Again the researcher would be expected to collect data and analyze the results. Applicants should have a strong background in statistical/neural network techniques and an interest in applying them to real world problems. The candidates will be experience in programming (MATLAB preferred) and mathematical analysis. Enquiries are welcome and can be made to Daniel Osorio or Roland Baddeley The salary will be on the standard BBSRC scale starting from 19,428 p.a. plus benefits. Applicants for this position should submit C.V., relevant reprints, and 3 letters of references. Letters of recommendation can be sent electronically (plain text, word, or PDF) to rjb at biols.susx.ac.uk Applications can also be sent by mail to: Roland Baddeley, Laboratory of Experimental Psychology, Sussex University, Falmer, Brighton England BN1 8QG. Replies will only be made to people who get called for interview. Closing date 30th October 2000. -- Dr Roland Baddeley Laboratory of Experimental Psychology Sussex University, Falmer Tel: (01273) 678961 Fax: (01273) 678611 WWW: http://www.biols.susx.ac.uk/Home/Roland_Baddeley/ From vaina at engc.bu.edu Mon Jul 31 15:59:32 2000 From: vaina at engc.bu.edu (Lucia M. Vaina) Date: Mon, 31 Jul 2000 15:59:32 -0400 Subject: T POSITION in fMRI studies of neuroplasticity--PLEASE POST!! Message-ID: Hi I would appreciate it if you could post this research position in my laboratory. Thank you. ******* The Brain and Vision Research Laboratory has an opening in the area of psychophysics and fMRI studies of adult cortical plasticity (stroke recovery and perceptual learning). An exciting new position is available for a full-time research assistant to work in Dr. Lucia Vaina's lab at Boston University http://www.bu.edu/eng/labs/bravi Our lab carries out computational and experimental research on the patterns of adult plasticity after extrastriate lesions (due to embolic stroke), perceptual learning and aspects of visual perception relevant to linking perception and action. We use psychophysics and fMRI to study both stroke patients and neurologically intact subjects. This job primarily involves assisting with fMRI experiments, including the preparation of stimulus materials, the running of subjects in an MRI scanner, the analysis of MRI data, and the preparation of this data in figures for publications and presentations. Requirements: 1. experience with multiple computer platforms ( Mac&PC, Microsoft Office, Unix, and Linux). Some knowledge of programming languages (e.g., Matlab and C) is desirable. 2. ability to trouble-shoot various hardware and software problems with the lab computers (e.g. printers, extension conflicts, network connections, etc). 3. excellent organizational and interpersonal skills. The position will involve maintaining the equipment and ordering supplies,conducting literature searches, and making appointments for subjects. 4. willingness to work on "odd hours" (depending on the magnet-time allocated to our group) and occasionally on late nights/weekends are both essential. 5. knowledge of and interest in visual cognition, neuroanatomy, and especially statistics is a great plus. This job is a good match for anyone interested in research in the neurology of vision or in cognitive neuroscience to learn more about these fields by becoming a member of a lively and productive interdisciplinary research group housed at multiple sites in Boston. *** Salary commensurable with experience*** The job can begin any time after September 1. Commitment of one year is required. Send inquiries, CV and names and phone numbers of 2-3 references to vaina at bu.edu or best to Professor Lucia M. Vaina Lucia M. Vaina Ph.D., Sc.D. Brain and Vision Research Laboratory Boston University, Department of Biomedical Engineering College of Engineering 44 Cummington str, Room 315 Boston University Boston, Ma 02215 USA tel: 617-353-2455 fax: 617-353-6766 From sshams at biodiscovery.com Mon Jul 31 12:14:31 2000 From: sshams at biodiscovery.com (Soheil Shams) Date: Mon, 31 Jul 2000 09:14:31 -0700 Subject: Job Posting: Senior Image Processing Scientist Message-ID: Senior Image Processing Scientist Description: We are looking for a talented, creative individual with a strong background in image processing, machine vision, and neural networks to direct and participate in the development of existing and new image processing products for genetic analysis. This position involves development and implementation of machine vision and image processing algorithms, such as image segmentation, multi-spectral image processing, and statistical and ANN based image quality assessment, customer interactions, research, and providing educational training services. The position requires the ability to formulate problem descriptions through interaction with end-user customers. These technical issues must then be transformed into innovative and practical algorithmic solutions. We expect our scientists to have outstanding written/oral communication skills and encourage publications and presentations in scientific journals and conferences. Requirements: - A Ph.D. in Electrical Engineering, Computer Science, or related field, or equivalent experience. - Experience developing image processing and pattern recognition algorithms. - Solid experience with Object Oriented programming in Java and/or C++ - Knowledge of biology and genetics is a plus but not necessary. - Excellent communication skills. For consideration, please send your r?sum? along with a cover letter to E-mail: HR at BioDiscovery.com Fax: (310) 966-9346 Snail-mail: BioDiscovery, Inc. 11150 W. Olympic Blvd. Suite 1170 Los Angeles, CA 90064 BioDiscovery, Inc. is start-up company founded in 1997 and dedicated to the development of state-of-the-art bioinformatics tools for biotechnology research applications. We are a leading gene expression image and data analysis firm with an outstanding client list and a progressive industry stance. BioDiscovery is an equal opportunity employer with good benefits, and our shop has a friendly, high-energy atmosphere. We are headquartered in sunny Southern California near the UCLA campus. From nschweighofer at neurotek.co.jp Mon Jul 31 02:37:09 2000 From: nschweighofer at neurotek.co.jp (Nicolas Schweighofer) Date: Mon, 31 Jul 2000 15:37:09 +0900 Subject: job offer in Tokyo Message-ID: <592E79EA9603D411B08F00D0B7203CB1053EC0@LCTKY01> My company has two research positions that could be of interest for people with training in statistics/neural networks. Learning Curve K.K, Tokyo, Japan: We have developed a new, scientific, and patented learning method, which has the potential to radically change the way we learn. Our company is now preparing to introduce this method to both the Japan and U.S. markets in the form of computer software applications, and we are expanding our research team for the development and testing of future versions of this method. The positions require a Masters or a PhD degree with training in computer science, applied math, physics, psychology, cognitive neuroscience, or related field. Depending on the position, the candidates should either have strong analytical skills preferably with experience in statistics/neural networks or have proven experience in designing and conducting experiments with subjects. Applicants should be creative, have the ability to quickly learn new techniques and concepts, and have excellent oral and written skills in either English or Japanese. We offer very competitive salaries and an employee Stock Option Plan. If you are looking for the opportunity to conduct high-level research in an exciting pre-IPO start-up in Tokyo, please send your resume to nicolas at neurotek.co.jp. Nicolas Schweighofer, Ph.D. Director of Core Research The Learning Curve, Inc. Tel: +81-3-3463-7266 Fax: +81-3-5489-7015 nicolas at neurotek.co.jp From iiass.alfredo at tin.it Mon Jul 3 13:26:43 2000 From: iiass.alfredo at tin.it (Alfredo Petrosino) Date: Mon, 03 Jul 2000 19:26:43 +0200 Subject: Neural Nets School 2000 Message-ID: <3960CCD3.65CD2D10@tin.it> 5th Course of International Summer School "Neural Nets E. R. Caianiello" on Visual Attention Mechanisms 23-29 October 2000 International Institute for Advanced Scientific Studies (IIASS) Vietri sul Mare, Salerno (Italy) http://www.iiass.it/nnschool JOINTLY ORGANIZED BY International Institute for Advanced Scientific Studies (IIASS) Ettore Majorana Foundation and Center for Scientific Culture (EMFCSC) SPONSORED BY Salerno University, Department of "Scienze Fisiche", Italy Gruppo Nazionale di Cibernetica e Biofisica del CNR Pavia University, Department of Electrical Engineering, Italy DIRECTOR OF THE 5TH COURSE Virginio CANTONI (Pavia University, Italy) DIRECTORS OF THE SCHOOL Michael JORDAN (University of California, Berkely, USA) Maria MARINARO (Salerno University, Italy) ORGANIZING COMMITEE Virginio CANTONI (Pavia University, Italy) Maria MARINARO (Salerno University, Italy) Alfredo PETROSINO (INFM-Salerno University, Italy) The school, open to all suitably qualified scientists from around the world, is organized in lectures, panel discussions and poster presentations and will cover a number of broad themes relevant to Visual Attention, among them: - Foundation: Early vision, Visual streams, Perception and action, Log-map analysis - Attentional mechanisms: Pop-out theory, Texton theory, Contour integration and closure, Fuzzy engagement mechanisms - Visual search : Attentional control, Selective attention, Spatial attention, Detection versus discrimination - Multiresolution and planning : Complexity of search tasks, Hierarchical perceptual loops, Multiresolution and associative memory systems, Attention and action planning - Attentional Visual Architectures: Neural models of visual attention, Hierarchical and associative networks, Attentional pyramidal neural mechanisms - Experiences: Eyeputer and scanpath recorders, etc. INVITED SPEAKERS (the list is not complete): Virginio CANTONI, Pavia University Leonardo CHELAZZI, Verona University, Italy Vito DI GESU`, Palermo University, Italy Hezy YESHURUN, Haifa University, Israel Zhaoping LI, Gatsby, University College, London, UK Luca LOMBARDI, Pavia University, Italy Carlo Alberto MARZI, Verona University, Italy Alain MERIGOT, University of Paris Sud, France Eliano PESSA, Roma University, Italy Alfredo PETROSINO, INFM-Salerno University, Italy Marco PIASTRA, Pavia University, Italy Vito ROBERTO, Udine University, Italy Dov SAGI, Weizmann University, Israel John TSOTSOS, Center for Computer Vision, Canada Daniela ZAMBARBIERI, Pavia University, Italy Harry WECHSLER, George Mason University, USA Steven YANTIS, Johns Hopkins University, USA POETIC TOUCH Vietri (from "Veteri", its ancient Roman name) sul Mare ("on sea") is located within walking distance from Salerno and marks the beginning of the Amalfi coast. Short rides take to Positano, Sorrento, Pompei, Herculaneum, Paestum, Vesuvius, or by boat, the islands of Capri, Ischia, and Procida. Velia (the ancient "Elea" of Zeno and Parmenide) is a hundred kilometers farther down along the coast. Visit http://www.comune.vietri-sul-mare.sa.it/ for more information. FEE The full school fee is 1200 dollars, reduced to 1000 dollars for students. The fee includes accommodations in twin room, meals, one day of excursion, and a copy of the proceedings of the school. A supplement of 40 dollars per night should be paid for single room. A restricted number of scholarships are also available for students which make the specific request. DATES Application deadline: 8 September 2000 Acceptance notification : 20 September 2000 For further information please contact : Dr. A.Petrosino Fax: + 39 89 761189 Email: iiass.alfredo at tin.it ============================CUT====================================== APPLICATION FORM Title:_______ Family Name: ________________________________________________________ Other Names:_________________________________________________________ Name to appear on badge: ____________________________________________ MAILING ADDRESS: Institution _________________________________________________________ Department __________________________________________________________ Address _____________________________________________________________ State ____________________________ Country __________________________ Phone:____________________________ Fax: _____________________________ E-mail: _____________________________________________________________ Arrival date: __________________ Departure date: ____________________ Will you be applying for a scholarship ? yes/no (Please include in your application the amount of bursary support and a justification for the request) Will you submit a poster ? yes/no (Please include a one page abstract for review by the organizers). ============================CUT====================================== Please send the application form by electronic mail to: iiass.alfredo at tin.it, subject: Neural Nets school; or by fax to: Neural Nets School, +39 89 761 189 or by ordinary mail to the address: Neural Nets School IIASS, Via Pellegrino 19 I84019 Vietri sul Mare (Sa) Italy From fritz at neuro.informatik.uni-ulm.de Tue Jul 4 09:15:52 2000 From: fritz at neuro.informatik.uni-ulm.de (Fritz Sommer) Date: Tue, 4 Jul 2000 15:15:52 +0200 (MET DST) Subject: No subject Message-ID: <200007041315.PAA07533@cerebellum.informatik.uni-ulm.de> cneuro at bbb.caltech.edu, bisc-group at cs.berkely.edu From kap-listman at wkap.nl Tue Jul 4 20:07:19 2000 From: kap-listman at wkap.nl (kap-listman@wkap.nl) Date: Wed, 05 Jul 2000 02:07:19 +0200 (MEST) Subject: New Issue: Neural Processing Letters. Vol. 11, Issue 3 Message-ID: <200007050007.CAA27310@wkap.nl> Kluwer ALERT, the free notification service from Kluwer Academic/PLENUM Publishers and Kluwer Law International ------------------------------------------------------------ Neural Processing Letters ISSN 1370-4621 http://www.wkap.nl/issuetoc.htm/1370-4621+11+3+2000 Vol. 11, Issue 3, June 2000. TITLE: A Batch Learning Vector Quantization Algorithm for Nearest Neighbour Classification AUTHOR(S): Sergio Bermejo, Joan Cabestany KEYWORD(S): Learning Vector Quantization, Newton-s optimization, nearest neighbour classification, batch learning algorithms. PAGE(S): 173-184 TITLE: A Comparison between the Tikhonov and the Bayesian Approaches to Calculate Regularisation Matrices AUTHOR(S): Andreu Catala, Cecilio Angulo KEYWORD(S): Bayesian inference, ill-posed problems, neural networks, RBF, regularization techniques, smoothing functions. PAGE(S): 185-195 TITLE: Competitive and Temporal Inhibition Structures with Spiking Neurons AUTHOR(S): E. Ros, F. J. Pelayo, P. Martin-Smith, I. Rojas, D. Palomar, A. Prieto KEYWORD(S): spiking neurons, competitive processing, temporal inhibition, attentional control mechanisms, bio-inspired neural systems. PAGE(S): 197-208 TITLE: An Experimental Comparison of Three PCA Neural Networks AUTHOR(S): Simone Fiori KEYWORD(S): principal component analysis, generalized Hebbian learning, adaptive principal-component extraction. PAGE(S): 209-218 TITLE: Neural Net Based Hybrid Modeling of the Methanol Synthesis Process AUTHOR(S): Primoz Potocnik, Igor Grabec, Marko Setinc, Janez Levec KEYWORD(S): hybrid modeling, genetic algorithms, feature selection, methanol synthesis, neural networks. PAGE(S): 219-228 TITLE: Transfusion Cost Containment for Abdominal Surgery with Neural Networks AUTHOR(S): Steven Walczak, John E. Scharf KEYWORD(S): neural networks, abdominal surgery, AAA, transfusion, cost, MSBOS. PAGE(S): 229-238 -------------------------------------------------------------- Thank you for your interest in Kluwer's books and journals. NORTH, CENTRAL AND SOUTH AMERICA Kluwer Academic Publishers Order Department, PO Box 358 Accord Station, Hingham, MA 02018-0358 USA Telephone (781) 871-6600 Fax (781) 681-9045 E-Mail: kluwer at wkap.com Kluwer Law International Order Department 675 Massachusetts Avenue Cambridge, MA 02139 USA Telephone: (617) 354-0140 Toll-free (US customers only): 800 577-8118 Fax: (617) 354-8595 E-mail: sales at kluwerlaw.com EUROPE, ASIA AND AFRICA Kluwer Academic Publishers Distribution Center PO Box 322 3300 AH Dordrecht The Netherlands Telephone 31-78-6392392 Fax 31-78-6546474 E-Mail: orderdept at wkap.nl From rfrench at ulg.ac.be Wed Jul 5 08:50:24 2000 From: rfrench at ulg.ac.be (Robert French) Date: Wed, 05 Jul 2000 14:50:24 +0200 Subject: POSTDOCTORAL AND Ph.D OPPORTUNITIES FOR EUROPEANS in connectionist neural network modelling Message-ID: <4.1.20000705133723.00ad24c0@pop3.mailst.ulg.ac.be> POSTDOCTORAL AND Ph.D OPPORTUNITIES FOR EUROPEANS in connectionist neural network modelling We have received a four-year grant from the European Commission grant to study the basic mechanisms of learning and forgetting in natural and artificial neural systems. The work will be done at five universities in England, Belgium and France and will be a highly multi-disciplinary effort involving experimental psychology, computer and mathematical modelling, and neural imaging techniques. There is funding for 5 fixed-term (1 to 2 years) post-doctoral research fellows, and 5 PhD studentships distributed across the participating institutions. Learning new information can potentially interfere severely with previously learned information. In order to prevent this from happening, many researchers now believe that the brain evolved a =93dual memory=94 architecture in which information is first processed in the hippocampus and thereafter gradually consolidated in the neo-cortex. But exactly how this works is still far from clear. We hope to gain a better understanding of the mechanisms and the implications of hippocampal-neocortical information transfer. Beyond developing a formal understanding of this system, the project will also explore applications of this type of memory architecture in the areas of child development, implicit learning, and various forms of aphasia and amnesia. This research has a number of important implications for understanding human memory. In particular, the dual-network connectionist architecture makes a number of rather unexpected predictions concerning the evolution over time of representations in long-term memory. One of the major goals of this project is to determine whether or not this representational evolution occurs in the brain. This work may also shed light on infant memory consolidation and category-specific deficits observed in certain types of aphasia. The project includes funding for 5 fixed-term (1 to 2 years) post-doctoral research fellows, and 5 PhD studentships. The focus of research for these positions will depend on the institution of appointment. Although there is considerable overlap between the expertise in the different partner institutions, in general, the research focus at each institution will be as follows: University of Li=E8ge, Belgium (with Bob French): investigations of memory consolidation and forgetting using experimental and modelling methodologies University of Grenoble, France (Bernard Ans/St=E9phane Rousset): investigations of transfer mechanisms between network systems and between hippocampus and neocortex using modeling and neural imaging methodologies University of Warwick, UK (Nick Chater): Formal analysis of data compression during information transfer between network systems Birkbeck College, UK (Denis Mareschal): implications of dual memory system for understanding infant and cognitive development using experimental and modelling methodologies Universit=E9 Libre de Bruxelles, Belgium (Axel Cleeremans): investigations of implicit learning and consciousness using experimental and modelling methodologies The program is highly multi-disciplinary with a particular emphasis on connectionist modelling. The members of the project will do experimental work as well as modelling and will be based in one of the five universities but will be expected to interact at a high level with the other teams involved in the project. For example, a student focusing on modelling at the University of Li=E8ge will interact with and visit the University of Grenoble to acquire neural imaging skills. Considerable emphasis will be placed on an exchange of ideas among project participants and publication in international journals. The ideal candidate will have good programming skills, a knowledge of connectionist models, experimental skills, an interest in human memory, and a willingness to work in an inter-disciplinary setting. The candidate must be aged 35 years or less at the time of appointment (excluding time spent for compulsory military service or childcare). They must be nationals of a European Community Member State or an Associated State* or must have resided in the community for the last 5 years prior to their appointment. They must not be nationals of the host country they are applying to and must not have carried out their normal activities in the host country for more than 12 of the 24 months prior to their appointment. Doctoral candidates must have completed their undergraduate degree and post-doctoral candidates must have completed their doctorate. Knowledge of the language of the host country is helpful, but is not necessary. The language of the project is English. Salaries for these positions vary but are competitive. The exact amounts are stipulated by the regulations of the participating universities. For more information on the exact value of the stipends, please contact the project director at the university that you are interested in attending. These positions are available immediately. There is no closing deadline for application. Applications will continue to be accepted until the positions are filled. Interested candidates should directly contact the following individuals at the participating institutions for further details: Bob French (rfrench at ulg.ac.be) at the University of Liege Denis Mareschal (d.mareschal at bbk.ac.uk) at Birkbeck College Nick Chater (nick.chater at warwick.ac.uk) at Warwick University Axel Cleeremans (axcleer at ulb.ac.be) at the Universite Libre de Bruxelles Bernard Ans (Bernard.Ans at upmf-grenoble.fr) at the University of Grenoble. If you have any questions concerning the project, feel free to contact the project director, Bob French (rfrench at ulg.ac.be). *European Union Associated States are: Bulgaria, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Slovakia, Slovenia, Lichtenstein, Iceland, Israel, and Norway. ---------------------------------------------------------------------------- Robert M. French, Ph.D Quantitative Psychology and Cognitive Science Psychology Department University of Liege 4000 Liege, Belgium Tel: (32.[0]4) 366.20.10 FAX: (32.[0]4) 366.28.59 email: rfrench at ulg.ac.be URL: http://www.fapse.ulg.ac.be/Lab/cogsci/rfrench.html ---------------------------------------------------------------------------- From edwin at arti.vub.ac.be Wed Jul 5 12:58:18 2000 From: edwin at arti.vub.ac.be (Edwin de Jong) Date: Wed, 5 Jul 2000 18:58:18 +0200 (MET DST) Subject: PhD thesis available: Autonomous Formation of Concepts and Communication Message-ID: <200007051658.SAA07270@arti6.vub.ac.be> Dear Connectionists, this is to announce the availability of my PhD thesis for download. URL: http://arti.vub.ac.be/~edwin/thesis Title: Autonomous Formation of Concepts and Communication Best regards, Edwin de Jong __ Keywords: concept formation, evaluative feedback, reinforcement learning, situation concepts, development of communication, dynamical systems. Abstract Autonomous agents receive sensor input from the environment, select actions, and may receive evaluative feedback on their behavior. Multiple agents that are present in the same environment can benefit from using communication to overcome uncertainty in their information about the environment. The research in the thesis addresses the question of how concepts about the environment can be formed, and how a system of communication can develop that allows agents to exchange information about their environment. Instead of assuming that concepts are already available or based on universal primitives, in the approach followed here they are formed in response to interaction with the environment. Evaluative feedback can play an important role in this. This distinguishes the approach from concept learning, which requires supervised feedback. A particular type of concepts is described, called situation concepts. Situation concepts consist of features in the history of interaction between the agent and its environment, and predict some aspect of the future evolution of the state of the environment, possibly conditioned on the actions the agent may take. Several existing methods, particularly from the field of reinforcement learning, can be viewed as constructing a form of situation concepts, and a particular method for constructing a specific type of situation concepts is described. Since situation concepts convey information about the environment, they are especially suited for use in communication. The development of communication consists of the formation of associations between words and the concepts formed by individual agents, and is viewed as the behavior of a dynamical system. The variables of this system are the strengths of the associations between words and the concepts of all agents in a population. An algorithm for association formation for individual agents is described that leads to a common system of communication. A deterministic version of the system is shown mathematically and demonstrated experimentally to have point attractors that correspond to perfect communication. The stochastic system system has points in its phase space that play a similar role, and is preferable since it avoids certain deadlock situations. This finding is confirmed by an investigation of the relation between the amount of stochasticity and the quality of communication. The research contributes to the view of the development of communication as the behavior of a dynamical system. Finally, systematic measures are provided for the quality of conceptual systems and communication systems that can be used when the subject of communication, called referent, is known. From sheri.mizumori at psych.utah.edu Wed Jul 5 17:54:06 2000 From: sheri.mizumori at psych.utah.edu (Sheri Mizumori) Date: Wed, 05 Jul 2000 15:54:06 -0600 Subject: behavioral neuroscience postdoc positions at the University of Washington Message-ID: <3963AE7E.77C4FDC8@csbs.utah.edu> Post Doctoral Positions in Behavioral Neuroscience Two NIH funded positions are available for research concerning the neurobiology of learning and memory. Recent research from this lab has investigated the relative contributions of multiple brain structures to adaptive navigation by rats. These and other ongoing studies involve parallel recordings of single unit activity within single brain as well as across diverse neural systems. Our goal is to understand how neural systems dynamically interact as a function of experience. In addition, we are interested in how these interactions change as a function of changing brain states such as that which occur during normal aging and pathological conditions. The neural dynamics mediating learning are also explored with computational models. The lab will be moving to the University of Washington in August, 2000. Positions can start anytime after our arrival at the UW. Interested persons should send a vita and names of 3 references to: Sheri J. Y. Mizumori Dept. Psychology 390 S. 1530 E. Rm. 502 Univ. Utah Salt Lake City, UT 84112 Office: 801-581-5555 FAX: 801-581-5841 Email: mizumori at psych.utah.edu After Aug. 1, send to: Sheri J. Y. Mizumori Dept. Psychology Box 351525 University of Washington Seattle, WA 98195-1525 From Rolf.Wuertz at neuroinformatik.ruhr-uni-bochum.de Fri Jul 7 11:33:03 2000 From: Rolf.Wuertz at neuroinformatik.ruhr-uni-bochum.de (Rolf Wuertz) Date: Fri, 7 Jul 2000 17:33:03 +0200 (MEST) Subject: JOB: Computer vision Message-ID: <200007071533.e67FX3918995@fsnif.neuroinformatik.ruhr-uni-bochum.de> The Chair of Systems Biophysics (Prof. C. von der Malsburg) at the Institute for Neurocomputing of the Ruhr-University of Bochum in Germany has vacancies for physicists, computer scientists, or electrical engineers or holders of other relevant degrees. We have a variety of exciting projects in computer vision, image understanding, and machine learning. Special emphasis is put on the study of mechanisms for learning solutions to hard vision problems from examples. The concrete projects will be tailored to match the candidates' skills and interests. The research group has been founded in 1990 and has a long record of successful projects in face recognition, gesture recognition, object recognition, robot vision, network self-organization, and models of the visual cortex. For details take a look at our WWW-site: http://www.neuroinformatik.ruhr-uni-bochum.de/ini/VDM/ Candidates should have a degree in one of the above subjects, a dedication to do active research, and an interest in computer vision. Analytical and programming skills are required, good knowledge of C++ will be an asset. Good students, who wish to pursue a diploma (Diplom) degree at our institute, are also encouraged to apply. The appointment will be for an initial period of 2 years. This can be extended, and a Ph.D. can be earned. Payment will be in accordance with BAT, the German salary scale for public employees. Please send your detailed CV including supporting material and statement of interests (email is OK) to: +----------------------------------+-----------------------------+ | Dr. Rolf P. Wuertz | Phone: +49 234 32-27994 | | Institut fuer Neuroinformatik | or: +49 234 32-27997 | | Ruhr-Universitaet Bochum | Fax: +49 234 32-14210 | | D-44780 Bochum, Germany | Room: ND03-32 | +----------------------------------+-----------------------------+ | http://www.neuroinformatik.ruhr-uni-bochum.de/ini/PEOPLE/rolf/ | | Rolf.Wuertz at neuroinformatik.ruhr-uni-bochum.de | +----------------------------------------------------------------+ From angelo at soc.plym.ac.uk Fri Jul 7 07:15:02 2000 From: angelo at soc.plym.ac.uk (Angelo Cangelosi) Date: Fri, 07 Jul 2000 12:15:02 +0100 Subject: PhD in Visualisation and Virtual Reality Message-ID: <3.0.6.32.20000707121502.007d8ca0@soc.plym.ac.uk> EPRSC PhD Studentship A PhD student is sought for an exciting and innovative project aimed at the visualisation of large data sets recorded by Neurophysiologists. The aim of this project is to develop a new approach to the analysis of experimental data in neurophysiology based on the use of computer science techniques such as graphical engineering, visualisation and virtual reality. This new approach will provide neuroscientists with an interactive environment within which to explore their data sets. The primary focus of this research is on the analysis of multi-dimensional data sets. By navigating through these large data sets, researchers will be able to focus on particular features of the data as well as identifying overall characteristics. This will support the integration of the experimental and theoretical approaches to the analysis of neurophysiological evidence. This three year studentship is funded by the UK Engineering and Physical Sciences Research Council. It is managed by Dr. L. Stuart of the Centre of Neural and Adaptive Sytsems in the School of Computing at the University of Plymouth. The student will be involved in the development of software to support the visualisation of vast quantities of neurophysiological data. Much of this development will draw on current software such as VTK and VisAD as well as the development of new representations incorporating 3D technology and/or Virtual Reality. The bursary for this studentship 6200 UK Pounds. Students are permitted to supervise/teach up to six hours per week to supplement this income. The student must have a high level of knowledge and experience of C++. Additionally, they must have good oral and written communication skills in the English language. Due to the nature of the project, the student should be able to work individually and as part of a team. For more information see http:\\www.tech.plym.ac.uk\soc\research\neural\staff\lstuart\vacancy.htm or contact Dr. L. Stuart at lstuart at plymouth.ac.uk Applications should be received before 14th April 2000, but the position will remain open until a suitable candidate is found. Candidates should send a CV, a description of motivations and the name and address of two referees to: Carole Watson School of Computing, University of Plymouth Drake Circus, Plymouth PL4 8AA, United kingdom Telephone: 01752 232541 Fax: 01752 232540 Email: c.watson at plymouth.ac.uk Thanks Liz =============================================== Dr. Liz Stuart Senior Lecturer, University of Plymouth Drake Circus, Plymouth Devon, PL4 8AA United Kingdom Telephone: +44 1752 232665 Secretary: +44 1752 232541 Fax: +44 1752 232540 =============================================== From heiko.wersing at hre-ftr.f.rd.honda.co.jp Mon Jul 10 11:41:18 2000 From: heiko.wersing at hre-ftr.f.rd.honda.co.jp (Heiko Wersing) Date: Mon, 10 Jul 2000 17:41:18 +0200 Subject: Paper and thesis available: Spatial feature binding and segmentation Message-ID: <3969EE9E.29400B1E@hre-ftr.f.rd.honda.co.jp> Dear Connectionists, I would like to announce the following recently accepted paper and my PhD Thesis on spatial feature binding and learning in competitive neural layer architectures. The paper and thesis can be downloaded from my homepage at http://www.techfak.uni-bielefeld.de/~hwersing/ Comments and questions are welcome. ----------------------------------------------------------------------- "A competitive layer model for feature binding and sensory segmentation." by Heiko Wersing, Jochen J. Steil, and Helge Ritter to appear in Neural Computation Abstract: We present a recurrent neural network for feature binding and sensory segmentation, the competitive layer model (CLM). The CLM uses topographically structured competitive and cooperative interactions in a layered network to partition a set of input features into salient groups. The dynamics is formulated within a standard additive recurrent network with linear threshold neurons. Contextual relations among features are coded by pairwise compatibilities which define an energy function to be minimized by the neural dynamics. Due to the usage of dynamical winner-take-all circuits the model gains more flexible response properties than spin models of segmentation by exploiting amplitude information in the grouping process. We prove analytic results on the convergence and stable attractors of the CLM, which generalize earlier results on winner-take-all networks, and incorporate deterministic annealing for robustness against local minima. The piecewise linear dynamics of the CLM allows a linear eigensubspace analysis which we use to analyze the dynamics of binding in conjunction with annealing. For the example of contour detection we show how the CLM can integrate figure-ground segmentation and grouping into a unified model. ------------------------------------------------------------------------ "Spatial feature binding and learning in competitive neural layer architectures" by Heiko Wersing, PhD Thesis, Faculty of Technology, University of Bielefeld, March 2000 Abstract: The goal of this thesis is to contribute to the understanding of feature binding processes by investigating an artificial neural network model for spatial feature binding, the competitive layer model (CLM). The CLM is a recurrent neural network, which employs topographically structured competitive and cooperative interactions in a system of neural layers to represent binding by the layer-wise coactivation of feature-representing neurons. This model is explored by means of mathematical analysis and simulations on artificial and challenging real-world data. The first chapter motivates the issue of feature binding as one of the important questions regarding our understanding of brain function. The second chapter gives an overview of the controversial scientific discussion of the binding problem and reviews different neural network model approaches to binding with a focus on their application in sensory segmentation and perceptual grouping. In the third chapter some new theoretical results on the stability of general linear threshold networks are established, which allow to operate the CLM in a mode of strong contextual modulation. The conditions provide a regime, where linear threshold networks are at the same time sensitive to small changes in their inputs and are capable of strong recurrent amplification without runaway activity. In chapter four the competitive layer model is introduced and its structure and dynamics are described. A deterministic annealing mechanism is introduced, which is compared to Potts-spin mean field models. The stability conditions of chapter three are applied to ensure convergence of the CLM, and additional conditions are proved, which guarantee the Winner-Take-All behaviour of the competitive columnar interactions. The grouping dynamics is characterized in relation to the lateral interactions by performing an eigensubspace analysis. Finally, the lateral coupling scheme of the CLM is generalized to general labeling problems and the chapter concludes with a comparison to other labeling approaches. Chapter five presents the application of the CLM to model a wide range of Gestalt-based perceptual grouping laws. First, the grouping of contours according to the principle of continuity is considered by using oriented edge elements as features. The results are discussed for real images, and an extension of the contour grouping approach to the task of cell segmentation is described. Other grouping principles which are considered include motion grouping, greyscale segmentation, and texture segmentation. In chapter six methods are presented to obtain appropriate lateral interactions for grouping and segmentation by supervised learning processes from manually labelled training patterns. The methods are evaluated on an artificial data set and cell images from fluorescence microscopy. A different learning approach is presented in Chapter seven, which aims at optimizing the lateral connection structure in order to improve the capabilities for solving complex constraint satisfaction problems. The proposed backtracking deterministic annealing method is interpreted as a heuristic approach to combine classical backtracking with neural deterministic annealing and successfully applied to the problem of complex tiling problems. The concluding chapter summarizes the main results and discusses possible future research directions. --------------------------------------------------------------------- -- Heiko Wersing Future Technology Research HONDA R&D EUROPE (DEUTSCHLAND) GmbH Carl-Legien-Str. 30 63073 Offenbach /Main Germany Tel.: +49-69-89011741 Fax: +49-69-89011749 e-mail: heiko.wersing at hre-ftr.f.rd.honda.co.jp From vogdrup at daimi.au.dk Mon Jul 10 03:46:25 2000 From: vogdrup at daimi.au.dk (Jakob Vogdrup Hansen) Date: Mon, 10 Jul 2000 09:46:25 +0200 Subject: PhD thesis available: Combining Predictors ... Message-ID: <200007100746.JAA10035@ppp.brics.dk> Dear Connectionists, Some people have had problems downloading my PhD thesis. I therefore give four different links to the PhD thesis in postscript and pdf format. The two last links are identical to the two first except some (unnecessary) pictures have been removed. http://www.daimi.au.dk/~vogdrup/diss.ps http://www.daimi.au.dk/~vogdrup/diss.pdf http://www.daimi.au.dk/~vogdrup/diss2.ps http://www.daimi.au.dk/~vogdrup/diss2.pdf Comments are welcome. regards, Jakob Title: Combining Predictors. Meta Machine Learning Methods and Bias/Variance & Ambiguity Decompositions Abstract: The most important theoretical tool in connection with machine learning is the bias/variance decomposition of error functions. Together with Tom Heskes, I have found the family of error functions with a natural bias/variance decomposition that has target independent variance. It is shown that no other group of error functions can be decomposed in the same way. An open problem in the machine learning community is thereby solved. The error functions are derived from the deviance measure on distributions in the one-parameter exponential family. It is therefore called the deviance error family. A bias/variance decomposition can also be viewed as an ambiguity decomposition for an ensemble method. The family of error functions with a natural bias/variance decomposition that has target independent variance can therefore be of use in connection with ensemble methods. The logarithmic opinion pool ensemble method has been developed together with Anders Krogh. It is based on the logarithmic opinion pool ambiguity decomposition using the Kullback-Leibler error function. It has been extended to the cross-validation logarithmic opinion pool ensemble method. The advantage of the cross-validation logarithmic opinion pool ensemble method is that it can use unlabeled data to estimate the generalization error, while it still uses the entire labeled example set for training. The cross-validation logarithmic opinion pool ensemble method is easily reformulated for another error function, as long as the error function has an ambiguity decomposition with target independent ambiguity. It is therefore possible to use the cross-validation ensemble method on all error functions in the deviance error family. -- Jakob V. Hansen Tlf: 86 750618 Rydevnget 87, 1. th. Kontor: B2.15 Lokal: (8942)3355 8210 Aarhus V E-mail: Vogdrup at daimi.au.dk From bengio at idiap.ch Mon Jul 10 03:26:25 2000 From: bengio at idiap.ch (Samy Bengio) Date: Mon, 10 Jul 2000 09:26:25 +0200 (MET DST) Subject: open PhD/Postdoc positions in machine learning Message-ID: Open positions for Ph.D. and Postdoctoral candidates in Machine Learning The Dalle Molle Institute for Perceptual Artificial Intelligence (IDIAP, http://www.idiap.ch) seeks qualified applicants for Ph.D. and Postdoc positions for its Machine Learning group. Many projects are currently open, such as: (a) Mixture of kernel methods such as Support Vector Machines (SVMs) and generative models such as Hidden Markov Models (HMMs) for discriminative speech processing. This work should initially build upon the preliminary works from Jaakkola and Haussler, such as, "Exploiting generative models in discriminative classifiers", in Advances in Neural Information Processing Systems 11: Proceedings of the 1998 Conference, M. Kearns, S. Solla, and D. Cohn, eds., MIT Press, 1999, pp. 487-493. (b) Several extensions of mixture and ensemble models: - Feature selection for Mixture of Experts, - Mixture of Support Vector Machines, - Mixture of binary classifiers for multiclass problems. (c) Fusion of multimodal systems at different levels: - at the score level, - for confidence intervals, - during the learning process. The ideal Ph.D. candidate should have good background in statistics, optimization, and computer science. The ideal Postdoc candidate should have strong background in statistical learning theory in general, including SVMs, neural networks, and mixture models. All applicants should be familiar with C/C++ programming under a Unix environment. Although IDIAP is located in the French part of Switzerland, English is the main working language at IDIAP. Free English and French courses are also provided. IDIAP is located in the town of Martigny (http://www.martigny.ch) in Valais, a scenic region in the south of Switzerland, surrounded by the highest mountains of Europe, and offering exciting recreational activities, including hiking, climbing and skiing, as well as varied cultural activities. It is within close proximity to Montreux (Jazz Festival) and Lausanne (EPFL, http://www.epfl.ch). Candidates should send their detailed CV to Dr. Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From Nigel.Goddard at ed.ac.uk Tue Jul 11 07:27:02 2000 From: Nigel.Goddard at ed.ac.uk (Nigel Goddard) Date: Tue, 11 Jul 2000 12:27:02 +0100 Subject: Research Programmer in Neuronal Simulation Message-ID: <396B0486.F80A2E1C@ed.ac.uk> RESEARCH PROGRAMMER IN NEURAL SIMULATION METHODS Intitute for Adaptive and Neural Computation University of Edinburgh The NEOSIM project, funded by NIMH and NSF, is developing simulation environments and associated software tools for modelling of brain processes. The project has computer science and neuroscience research goals, and aims to develop distributable, portable, parallel software for the computational neuroscience community. We seek a Research Programmer to augment the existing multidisciplinary, multinational team. Some travel to EU countries or the US will be required. The Research Programmer will be involved in design, documentation, implementation, testing, dissemination, maintenance and development of the NEOSIM framework and will also be responsible for the installation and maintenance of software packages on the research computers, and may be involved in the specification, installation and system administration of research computers including small parallel platforms. Independent research related to the goals of the NEOSIM project, including modeling, will be encouraged. Tenable for up to 3 years subject to annual renewal and further renewal beyond 3 years subject to success in attracting further funding. Interested parties should contact me as soon as possible via email to Nigel.Goddard at ed.ac.uk. I will be at CNS2000 in Brugge. Sites: http://anc.ed.ac.uk/neosim http://www.informatics.ed.ac.uk -- ============================================== Dr. Nigel Goddard Institute for Adaptive and Neural Computation Division of Informatics University of Edinburgh 5 Forrest Hill Edinburgh EH1 2QL Scotland Telephone: +44 131 650 3087 email: Nigel.Goddard at ed.ac.uk web: http://anc.ed.ac.uk/~ngoddard Fax (paper): +44 131 650 6899 eFAX (email): +1 603 698 5854 ============================================== From pelletl at CRM.UMontreal.CA Tue Jul 11 10:36:22 2000 From: pelletl at CRM.UMontreal.CA (Louis Pelletier) Date: Tue, 11 Jul 2000 10:36:22 -0400 Subject: Workshop on MEMORY, DELAYS AND MULTISTABILITY Message-ID: ************ SECOND ANNOUNCEMENT - PLEASE CIRCULATE *************** CALL FOR PAPERS As part of the activities of the 2000-2001 Theme Year in Mathematical Methods in Biology and Medicine, the Centre de Recherches Mathematiques (CRM) of the Universite de Montreal is organizing an international workshop on "MEMORY, DELAYS AND MULTISTABILITY IN NEURAL SYSTEMS" MONTREAL, 12-15 October 2000 This workshop will focus on recent advances in the mathematical modeling of neural systems having one or more of the following dynamical features: 1) memory - in particular, "dynamic" memory, 2) synaptic and propagation delays, or 3) the coexistence of steady states. These features, as well as their possible interactions, have been highlighted in a number of recent studies, e.g. of the activity of recurrent neural circuits ubiquitous in the nervous system. The emphasis will be mainly on the mathematical modeling of real biological systems, yet the workshop will also explore the relevance of these issues to disciplines ranging from psychology to artificial neural networks. MORE INFORMATION AT: http://www.crm.umontreal.ca/biomath CONTRIBUTED PAPERS BEFORE 31 AUGUST 2000: The workshop will consist of invited talks (see below) as well as short oral and poster presentations. If you wish to contribute to the workshop, your title and abstract must be sent to us before 31 August 2000 for inclusion in the programme. Your abstract should not exceed 500 words. It should be submitted electronically in either text (ASCII), Latex, or Word format to the local workshop coordinator Louis Pelletier (PELLETL at crm.umontreal.ca) with the title of the workshop in the subject heading. Please indicate your presentation format preference (oral vs poster). REGISTRATION: Registration for the conference should be done no later than 15 September 2000 for attendance and accommodation, and 30 September 2000 for attendance only. Additional information on the CRM and accommodation as well as a registration form are available at http://www.crm.umontreal.ca/biomath. Questions on the scientific aspects of the workshop should be sent to the scientific organizer at alongtin at physics.uottawa.ca (Andre Longtin). All other questions, including those concerning accommodations, should be directed to Louis Pelletier (PELLETL at crm.umontreal.ca; 514-343-2197). INVITED SPEAKERS: K. Aihara, Mathematical Engineering and Information Physics, U. Tokyo S. Becker, Psychology, McMaster U., CAN P. Bressloff, Mathematics, U. Loughborough, UK N. Brunel, Labo. de Physique Statistique, ENS, Paris S.A. Campbell, Mathematics, U. Waterloo, CAN C. Canavier, Psychology, U. New Orleans, USA G.A. Carpenter, Cognitive and Neural Systems, Boston U., USA A. Destexhe, Physiology, Laval U., CAN M. Ding, Center for Complex Systems and Brain Sciences, Florida A.U., USA W. Gerstner, Computer Science, SFIT, Lausanne L. Glass, Physiology, McGill U., CAN (*) J. Guckenheimer, Center for Applied Math., Cornell, USA A.V.M. Herz, Theoretical Biology, Humboldt, Berlin E. Izhikevich, The Neurosciences Institute, La Jolla, USA W. Maass, Tech. Univ. Graz, Austria J.G. Milton, Neurology, U. Chicago K. Pakdaman, INSERM, Paris K. Pawelzik, Institute for Theoretical Physics, U. Bremen (*)J. Rinzel, Center for Neural Science, New York U., USA M. Tsodyks, Dept. of Neurobiology, Weizmann Inst., Israel X.-J. Wang, Volen Center, Brandeis U., USA (*=tentative) From stefan.wermter at sunderland.ac.uk Thu Jul 13 03:25:03 2000 From: stefan.wermter at sunderland.ac.uk (Stefan.Wermter) Date: Thu, 13 Jul 2000 08:25:03 +0100 Subject: Intl workshop on neural networks and neuroscience Message-ID: <396D6ECE.4D067733@sunderland.ac.uk> EmerNet3: Emerging Computational Neural Network Architectures based on Neuroscience International EPSRC Workshop Date: 8-9 August 2000 Location: Durham Castle, Durham, United Kingdom Workshop web page with current information is http://www.his.sunderland.ac.uk/worksh3 Organising Committee ----------------------- Prof. Stefan Wermter Chair Hybrid Intelligent Systems Group Informatics Centre, SCET University of Sunderland Prof. Jim Austin Advanced Computer Architecture Group Department of Computer Science University of York Prof. David Willshaw Institute for Adaptive and Neural Computation Division of Informatics University of Edinburgh -------------- Contact Details --------------- Mark Elshaw (Workshop Organization) Hybrid Intelligent Systems Group Informatics Centre, SCET University of Sunderland St Peter's Way Sunderland SR6 0DD United Kingdom Phone: +44 191 515 3249 Fax: +44 191 515 2781 E-mail: Mark.Elshaw at sunderland.ac.uk Prof. Stefan Wermter (Chair) Informatics Centre, SCET University of Sunderland St Peter's Way Sunderland SR6 0DD United Kingdom Phone: +44 191 515 3279 Fax: +44 191 515 2781 E-mail: Stefan.Wermter at sunderland.ac.uk http://www.his.sunderland.ac.uk/~cs0stw/ http://www.his.sunderland.ac.uk/ From hadley at cs.sfu.ca Tue Jul 18 17:40:37 2000 From: hadley at cs.sfu.ca (Bob Hadley) Date: Tue, 18 Jul 2000 14:40:37 -0700 (PDT) Subject: Systematicity, Hebbian Learning, Lang. Acquis. In-Reply-To: <199801152116.NAA10281@css.cs.sfu.ca> from "Bob Hadley" at Jan 15, 1998 01:16:40 PM Message-ID: <200007182140.OAA10206@css.css.sfu.ca> The following paper is available at: www.cs.sfu.ca/~hadley/online.html Total pages: 37 at 1.6 spacing. Syntactic Systematicity Arising from Semantic Predictions in a Hebbian-Competitive Network BY Robert F. Hadley Adam Rotaru-Varga Dirk V. Arnold Vlad C. Cardei School of Computing Science, Simon Fraser Burnaby, B.C., V5A 1S6 Canada ABSTRACT A Hebbian-inspired, competitive network is presented which learns to predict the typical semantic features of denoting terms in simple and moderately complex sentences. In addition, the network learns to predict the appearance of syntactically key words, such as prepositions and relative pronouns. Importantly, as a by-product of the network's semantic training, a strong form of syntactic systematicity emerges. This systematicity is exhibited even at a novel, deeper level of clausal embedding. All network training is unsupervised with respect to error feedback. A novel variant of competitive learning, and an unusual hierarchical architecture are presented. The relationship of this work to issues raised by Marcus (1998) and Phillips (2000) is explored. Keywords: Systematicity, Semantic Features, Language Acquisition, Competitive Learning, Connectionism. From mzorzi at ux1.unipd.it Mon Jul 17 17:18:48 2000 From: mzorzi at ux1.unipd.it (Marco Zorzi) Date: Mon, 17 Jul 2000 23:18:48 +0200 Subject: postdoc position in connectionist modelling Message-ID: <4.3.1.0.20000717231701.00b30810@ux1.unipd.it> POST-DOCTORAL FELLOWSHIP A two-years postdoctoral position in connectionist modelling is available with Dr. Marco Zorzi and Prof. Carlo Umilte at University of Padova, Department of Psychology. The project is part of a EU-funded Research Training Network of six sites (University College London, INSERM U334 Orsay, Universite Catholique de Louvain, University of Innsbruck, University of Trieste, and University of Padova) investigating "Mathematics and the Brain". The Padova team will focus on the computational bases of mathematical cognition (basic numerical abilities and simple arithmetic). The successful applicant will have a quantitative background and excellent programming skills (MATLAB/Visual Basic/Visual C). Specific experience and/or publications in neural networks/connectionist modelling is desirable. A background in cognitive neuroscience will be a plus but is not necessary. The international network offers an excellent opportunity for gaining experience in a wide range of methodologies in cognitive neuroscience. Short visits and exchanges between laboratories are also planned. Salary and benefits are highly competitive (35000 Euro per year). Applicants must be citizens of a EU country (excluding Italy), EFTA-EEA states, Candidate States, or Israel. Application deadline is 30 September 2000, start date is negotiable (after November 1st). Send CV, reprints or preprints and the names of two references (preferably by email) to: Dr. Marco Zorzi Prof. Carlo Umilt? Dipartimento di Psicologia Generale Universit? di Padova Via Venezia 8 35131 Padova (Italy) mzorzi at ux1.unipd.it umilta at ux1.unipd.it Dr. Marco Zorzi Dipartimento di Psicologia Generale Universit? di Padova via Venezia 8 35131 Padova tel: +39 049 8276635 fax: +39 049 8276600 email: mzorzi at psico.unipd.it (and) Institute of Cognitive Neuroscience voice: +44 171 3911151 University College London fax : +44 171 8132835 17 Queen Square London WC1N 3AR (UK) http://www.psychol.ucl.ac.uk/marco.zorzi/marco.html From swatanab at pi.titech.ac.jp Wed Jul 19 01:28:32 2000 From: swatanab at pi.titech.ac.jp (Sumio Watanabe) Date: Wed, 19 Jul 2000 14:28:32 +0900 Subject: Paper : Learning theory and algebraic analysis and geometry Message-ID: <002901bff142$2e0e7c20$918a7083@deskpowerts> Dear connectionists, I would like to announce the following paper accepted for publication in Neural Computation. This paper firstly clarifies the algebraic geometrical structure of the hierarchical learning machines. Questions and comments are welcome. +++++ Title: Algebraic analysis for non-identifiable learning machines. Cite: http://watanabe-www.pi.titech.ac.jp/~swatanab/appendix.html ABSTRACT: This paper clarifies learning efficiency of a non-identifiable learning machine such as a multi-layer neural network whose true parameter set is an analytic set with singular points. By using a concept in algebraic analysis, we rigorously prove that the free energy or the Bayesian stochastic complexity is asymptotically equal to $\lambda_{1}\log n -(m_{1}-1)\log\log n + $constant, where $\lambda_{1}$ is a rational number, $m_{1}$ is a natural number, and $n$ is the number of training samples. Also we show an algorithm to calculate $\lambda_{1}$ and $m_{1}$ based on the resolution of singularities in algebraic geometry. In regular models, $2\lambda_{1}$ is equal to the number of parameters and $m_{1}=1$, whereas in non-regular models such as mutilayer networks $2\lambda_{1}$ is not larger than the number of parameters and $m_{1}\geq 1$. Since the increase of the free energy is equal to the generalization error, the non-identifiable learning machines are the better models than the regular ones if Bayesian or ensemble training is applied. +++++ Sincerely, Dr. Sumio Watanabe, Associate Professor Advanced Information Processing Division Precision and Intelligence Laboratory Tokyo Institute of Technology E-mail: swatanab at pi.titech.ac.jp (Fax)+81-45-924-5018 From ingber at ingber.com Fri Jul 21 21:30:06 2000 From: ingber at ingber.com (Lester Ingber) Date: Fri, 21 Jul 2000 20:30:06 -0500 Subject: Paper: Optimization of Trading Physics Models of Markets Message-ID: <20000721203006.A15767@ingber.com> The following preprint is available: %A L. Ingber %A R.P. Mondescu %T Optimization of Trading Physics Models of Markets %D 2001 %J IEEE Trans. Neural Networks %O Invited paper for special issue on Neural Networks in Financial Engineering. URL http://www.ingber.com/markets01_optim_trading.ps.gz ABSTRACT We describe an end-to-end real-time S&P futures trading system. Inner-shell stochastic nonlinear dynamic models are developed, and Canonical Momenta Indicators (CMI) are derived from a fitted Lagrangian used by outer-shell trading models dependent on these indicators. Recursive and adaptive optimization using Adaptive Simulated Annealing (ASA) is used for fitting parameters shared across these shells of dynamic and -- Lester Ingber http://www.ingber.com/ PO Box 06440 Wacker Dr PO Sears Tower Chicago IL 60606-0440 http://www.alumni.caltech.edu/~ingber/ From terry at salk.edu Fri Jul 21 19:50:19 2000 From: terry at salk.edu (Terry Sejnowski) Date: Fri, 21 Jul 2000 16:50:19 -0700 (PDT) Subject: NEURAL COMPUTATION 12:8 Message-ID: <200007212350.QAA03312@dax.salk.edu> Neural Computation - Contents - Volume 12, Number 8 - August 1, 2000 ARTICLE Neural Systems as Nonlinear Filters Wolfgang Maass and Eduardo D. Sontag NOTE Analytical Model for the Effects of Learning on Spike Count Distributions Gianni Settanni and Alessandro Treves LETTERS Calculation of Interspike Intervals for Integrate-and-Fire Neurons with Poisson Distribution of Synaptic Inputs A. N. Burkitt and G. M. Clark Statistical Procedures for Spatio-Temporal Neuronal Data with Applications to Optical Recording of the Auditory Cortex O. Francois, L. Mohamed Abdallahi, J. Horikawa, I Taniguchi and T. Herve Probabilistic Motion Estimation Based on Temporal Coherence Pierre-Yves Burgi, Alan L. Yuille, and Norberto M. Grzywacz Boosting Neural Networks Holger Schwenk and Yoshua Bengio Gradient-based Optimization of Hyper-parameters Yoshua Bengio A Signal-Flow-Graph Approach to On-line Gradient Calculation Paolo Campolucci, Aurelio Uncini and Francesco Piazza Bootstrapping Neural Networks Jurgen Franke and Michael H. Neumann Information Geometry of Mean-Field Approximation Toshiyuki Tanaka Measuring the VC-Dimension Using Optimized Experimental Design Xuhui Shao, Vladimir Cherkassky and William Li ----- ON-LINE - http://neco.mitpress.org/ SUBSCRIPTIONS - 2000 - VOLUME 12 - 12 ISSUES USA Canada* Other Countries Student/Retired $60 $64.20 $108 Individual $88 $94.16 $136 Institution $430 $460.10 $478 * includes 7% GST MIT Press Journals, 5 Cambridge Center, Cambridge, MA 02142-9902. Tel: (617) 253-2889 FAX: (617) 258-6779 mitpress-orders at mit.edu ----- From bengio at idiap.ch Wed Jul 26 09:33:56 2000 From: bengio at idiap.ch (Samy Bengio) Date: Wed, 26 Jul 2000 15:33:56 +0200 (MET DST) Subject: Open positions in speech, computer vision, machine learning, and multimodal interfaces Message-ID: SEVERAL OPEN POSITIONS IN SPEECH, COMPUTER VISION, MACHINE LEARNING, AND MULTIMODAL INTERFACES In view of a massive growth of the Institute and the development of new activites oriented towards advanced multimodal interfaces, the Dalle Molle Institute for Perceptual Artificial Intelligence (IDIAP, http://www.idiap.ch), IDIAP currently welcomes applications of talented candidates at all levels with expertise or strong interest in the fields of speech processing, computer vision, machine learning, and multimodal interaction. The open positions include: management and senior positions (including one scientific deputy director and one speech processing group leader), project leaders, postdocs, and PhD students. It is expected that a few tenure track positions at the Assistant Professor level could also become available. Preference will be given to candidates with experience in one or several of the following areas: signal processing, statistical pattern recognition (typically applied to speech and scene analysis), neural networks, hidden Markov models, speech and speaker recognition, computer vision, human/computer interaction (dialog). Senior and postdoc candidates should also have a proven record of high quality research and publications. All applicants should be experienced in C/C++ programming and familiar with the Unix environment; they should also be able to speak and write in English (and be willing to learn French). ABOUT IDIAP IDIAP is a semi-private research institute affiliated with the Swiss Federal Institute of Technology at Lausanne (EPFL) and the University of Geneva. Located in Martigny (Valais, CH), IDIAP is partly funded by the Swiss Federal Government, the State of Valais, and the City of Martigny, and is involved in numerous national and international (European) projects. IDIAP is mainly carrying research and development in the fields of speech and speaker recognition, computer vision, and machine learning, and is recognized as a university level research laboratory, involving permanent senior scientists, postdocs, and PhD students (usually awarded an EPFL degree). IDIAP currently numbers around 35-40 scientists. LOCATION IDIAP is located in the town of Martigny (http://www.martigny.ch) in Valais, a scenic region in the South of Switzerland, surrounded by the highest mountains of Europe, and offering exciting recreational activities (including hiking, climbing and skiing), as well as varied cultural activities. It is also within close proximity to Montreux, Lausanne (EPFL) and Lake Geneva, and centrally located for travel to other parts of Europe. PROSPECTIVE CANDIDATES should send their detailed CV to: Prof. Herve Bourlard Director of IDIAP P.O. Box 592, Simplon, 4 CH-1920 Martigny, Switzerland Email: bourlard at idiap.ch Phone: +41-27-721.77.20; Fax: +41-27-721.77.12 ----- Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From bengio at idiap.ch Tue Jul 25 08:49:37 2000 From: bengio at idiap.ch (Samy Bengio) Date: Tue, 25 Jul 2000 14:49:37 +0200 (MET DST) Subject: SVMTorch: A new SVM program for Large-Scale Regression and Classification Problems Message-ID: I would like to inform you of the following new SVM software for large-scale regression and classification problems, available at http://www.idiap.ch/learning/SVMTorch.html. Information about this new software follows: SVMTorch A Support Vector Machine for Large-Scale Regression and Classification Problems Ronan Collobert (collober at idiap.ch) IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland Description SVMTorch is a new implementation of Vapnik's Support Vector Machine that works both for classification and regression problems, and that has been specifically tailored for large-scale problems (such as more than 20000 examples, even for input dimensions higher than 100). Source Code The source code is free for academic use. It must not be modified or distributed without prior permission the author. When using SVMTorch in your scientific work, please cite the following article: Ronan Collobert and Samy Bengio, Support Vector Machines for Large-Scale Regression Problems, IDIAP-RR-00-17, 2000. (available at ftp://ftp.idiap.ch/pub/reports/2000/rr00-17.ps.gz). The software has been successfully compiled on Sun/SOLARIS, Intel/LINUX and Alpha/OSF operating systems. Your can download it from ftp://ftp.idiap.ch/pub/learning/SVMTorch.tgz. Try It !!! First, you should download the source code from ftp://ftp.idiap.ch/pub/learning/SVMTorch.tgz and the examples from ftp://ftp.idiap.ch/pub/learning/TrainData.tgz. Put this two archive files in the same directory, and decompress them with zcat SVMTorch.tgz | tar xf - zcat TrainData.tgz | tar xf - It creates two new directories : "SVMTorch" and "TrainData". Now, go in the "SVMTorch" directory and edit the Makefile. You should only have to change the following lines, depending on your specific platform : # C-compiler #CC=gcc CC=cc # C-Compiler flags #CFLAGS=-Wall -W -O9 -funroll-all-loops -finline -fomit-frame-pointer -ffast-math CFLAGS=-native -fast -xO5 # linker #LD=gcc LD=cc # linker flags #LFLAGS=-Wall -W -O9 -funroll-all-loops -finline -fomit-frame-pointer -ffast-math LFLAGS=-native -fast -xO5 # libraries LIBS=-lm The default configuration is set for a machine running with the Sun Workshop compiler. An alternate (commented) configuration is proposed for the GNU gcc compiler. Type "make all" and pray. It should compile without any warning. For some platform, you could have to change the include files needed for "times", a non-standard function used by svm_torch. You would have to edit the file "general.h" and change the lines #ifdef I_WANT_TIME #include /*#include */ #include #endif If it doesn't work or if you don't want to measure the time of the learning machine, just comment the line : #define I_WANT_TIME Note that in "general.h" you can comment the line #define USEDOUBLE in order to do the computations in float. IT'S A BAD IDEA : svm_torch needs precision. If everything went well, you should have two programs : "svm_torch" and "svm_test". The first one is the learning machine and the second one is the testing machine. If you want to show all the options, just run svm_torch or svm_test without any parameter. To test the program in classification, try : svm_torch -v -ae ../TrainData/classif_train.dat ../TrainData/model_dummy It takes less than two minutes on a 300Mhz computer. You should have around 914 support vectors (this number could slightly change depending on the precision of your machine). To test the SVM on the train data, try : svm_test -ae ../TrainData/model_dummy ../TrainData/classif_train.dat You should have around 0.78% missclassified. To test the program in regression, try : svm_torch -v -ae -rm -st 900 -eps 20 ../TrainData/regress_train.dat ./TrainData/model_dummy You should have around 597 support vectors. Test the model with : svm_test -ae ../TrainData/model_dummy ../TrainData/regress_train.dat The mean squared error should be around 187.2. Options The general syntax of svm_torch and svm_test is svm_torch [options] example_file model_file svm_test [options] model_file test_file Where "example_file" is your training set file, "test_file" is your testing set file and "model_file" is the SVM-model created by svm_torch. All options are described when you launch svm_torch or svm_test without any argument. By default, svm_torch is a classification machine. If you want the regression machine, use option -rm. You should always use option -v with svm_torch : it gives a current error during learning. This error is only an indicator. It can oscillate. File format There are two main input formats for "input_file" and "test_file" in SVMTorch : an ASCII format, and a binary one. The ASCII format is the following: .... . . . .... where is an ASCII floating point number corresponding to the j-th value of the i-th example and is the i-th desired output (in classification, it should be +1/-1). With the same notation, the binary format is: ... ....... ... ... (First save the input table, then the output table, all in binary) There is another special input format for svm_test, when you don't have the desired output. (To use with the -no option). The ASCII version of this format is : .... . . . .... And the binary version is : ... ....... ... ----- Samy Bengio Research Director. Machine Learning Group Leader. IDIAP, CP 592, rue du Simplon 4, 1920 Martigny, Switzerland. tel: +41 27 721 77 39, fax: +41 27 721 77 12. mailto:bengio at idiap.ch, http://www.idiap.ch/~bengio From giro at cortex.hut.fi Tue Jul 25 04:43:00 2000 From: giro at cortex.hut.fi (Mark Girolami) Date: Tue, 25 Jul 2000 11:43:00 +0300 (EEST) Subject: PhD Studentship Available Message-ID: PhD Studentship Available Applied Computational Intelligence Research Unit University of Paisley Scotland, UK. & Power Technology Research Centre POWERGEN (UK) Nottingham, UK. Title of Investigation: 'The Discrimination of Novel Complex-Normal Behaviour from Faulty Operating Conditions for Large Rotodynamic Machines' A three year fully funded Phd studentship is available in the Applied Computational Intelligence Research Unit at the University of Paisley. This is a joint research project with POWERGEN (UK) and it is envisaged that the student will work over certain periods of time at the Power Technology Centre in Nottingham. Project Brief: The automated discrimination of genuine fault conditions from acceptable operating behaviour of large and complex rotodynamic machines is a challenging engineering problem. This is effectively a classification task, however as the available training samples consist largely of acceptable operating conditions and the occurrence of genuine fault conditions is relatively rare we have in effect to consider the problem of novelty detection. Standard classification techniques are not suitable for this particular problem as faults will be potentially diverse and ill sampled. There are also multi-state situations when a machine can be in a number of distinct states depending on operating history. There have been attempts in the past to use regression and neural network methods to try to model the relationship between machine vibration and operating conditions, but with limited success. The development and application of support vector/kernel methods to this problem domain is proposed as a potential avenue of investigation. Applications from prospective candidates with backgrounds in either mathematics, statistics, physics, engineering are especially welcome. Informal Enquiries can be made to: Prof. M.A. Girolami Chair of Computational Intelligence Department of Computing and Information Systems University of Paisley High Street Paisley, PA1 2BE Scotland, UK. e-mail: mark.girolami at paisley.ac.uk From giro at james.hut.fi Tue Jul 25 04:35:46 2000 From: giro at james.hut.fi (Mark Girolami) Date: Tue, 25 Jul 2000 11:35:46 +0300 (EEST) Subject: Advances in Independent Component Analysis Message-ID: New Book Now Available: Title: Advances in Independent Component Analysis Editor : Mark Girolami Publisher : Springer Verlag ISBN 1-85233-263-8 Companion Website : http://www.dcs.ex.ac.uk/ica Since 1995 the neural computing research community has produced a large number of publications dedicated to the Blind Separation of Sources (BSS) or Independent Component Analysis (ICA). This book is the outcome of a workshop which was held after the 1999 International Conference on Artificial Neural Networks (ICANN) in Edinburgh, Scotland. Some of the most active and productive neural computing researchers gathered to present and share new ideas, models and experimental results and this volume documents the individual presentations. Individual Chapter Titles 1. Hidden Markov Independent Component Analysis Authors : W. Penny, R.Everson, S. Roberts 2. Particle Filters for Non-Stationary ICA Authors : R. Everson, S.Roberts 3. Analysing the Independence of the Components by Topography Authors : A.Hyvarinen, P. Hoyer, M. Inki. 4. Dependent Component Analysis Author : A Barros 5. Ensemble Learning Authors : H. Lappalainen, J. Misken 6. Bayesian Non-linear ICA by MLP's Authors: H. Lappalainen, A. Honkela. 7. Ensemble Learning for Blind Separation and Deconvolution Authors : J.Misken, D. MacKay. 8. MUCICA for Rank-Deficient Distributions Authors : F. Palmieri, A.Budillon. 9. Blind Separation of Noisy image Mixtures Author : L.K. Hansen. 10. Searching for independence in Electromagnetic Brain Waves Authors : R.Vigario, J. Sarela, E. Oja. 11. ICA on Noisy Data : A Factor Analysis Approach Author : S Ikeda. 12. Analysis of Optical Imaging Data Using Weak Models and ICA Authors :J. Porril, J. Stone, J. Berwick, J. Mayhew, P. Coffey. 13. Independent Components in Text. Authors : T. Kolenda, L. Hansen, S.Sigurdsson. 14. Seeking Independence Using Biologically-Inspired ANN's Authors : P.Lai, D. Charles, C. Fyfe. From ted.carnevale at yale.edu Sun Jul 23 17:28:18 2000 From: ted.carnevale at yale.edu (Ted Carnevale) Date: Sun, 23 Jul 2000 17:28:18 -0400 Subject: NEURON course at SFN 2000 meeting Message-ID: <397B6372.515185E1@yale.edu> Short Course Announcement USING THE NEURON SIMULATION ENVIRONMENT Satellite Symposium, Society for Neuroscience Meeting 9 AM - 5 PM on Saturday, Nov. 4, 2000 Speakers: N.T. Carnevale, M.L. Hines, J.W. Moore, and G.M. Shepherd This 1 day course with lectures and live demonstrations will present information essential for teaching and research applications of NEURON, an advanced simulation environment that handles realistic models of biophysical mechanisms, individual neurons, and networks of cells. The emphasis is on practical issues that are key to the most productive use of this powerful and convenient modeling tool. Features that will be covered include: importing detailed morphometric data constructing and managing models of neurons with the CellBuilder constructing and managing network models with the new Network Builder using the Multiple Run Fitter to optimize models that have high-dimensional datasets NEURON's new functions for parallelizing simulations over a workstation cluster (32-bit MSWindows and/or UNIX/Linux boxes) database resources for empirically-based modeling Each registrant will receive a CD-ROM with software, plus a comprehensive set of notes that includes material which has not appeared elsewhere in print. For more information see the course's WWW pages at http://www.neuron.yale.edu/easy2k.html --Ted Supported in part by the National Science Foundation. Opinions expressed are those of the authors and not necessarily those of the Foundation. From pelillo at dsi.unive.it Fri Jul 21 03:28:28 2000 From: pelillo at dsi.unive.it (Marcello Pelillo) Date: Fri, 21 Jul 2000 09:28:28 +0200 (MET DST) Subject: An IEEE-TNN paper on Max-Weight-Clique In-Reply-To: Message-ID: Dear all, the following paper, accepted for publication in the IEEE Transactions on Neural Networks, is accessible at the following www site: http://www.dsi.unive.it/~pelillo/papers/ieee-tnn-mwcp.ps.gz Comments and suggestions are welcome! Best regards, Marcello Pelillo ********************* Approximating the Maximum Weight Clique Using Replicator Dynamics I. M. Bomze (University of Vienna, Austria) M. Pelillo (University of Venice, Italy) V. Stix (University of Vienna, Austria) Abstract Given an undirected graph with weights on the vertices, the maximum weight clique problem (MWCP) is to find a subset of mutually adjacent vertices (i.e., a clique) having largest total weight. This is a generalization of the classical problem of finding the maximum cardinality clique of an unweighted graph, which arises as a special case of the MWCP when all the weights associated to the vertices are equal. The problem is known to be NP-hard for arbitrary graphs and, according to recent theoretical results, so is the problem of approximating it within a constant factor. Although there has recently been much interest around neural network algorithms for the unweighted maximum clique problem, no effort has been directed so far towards its weighted counterpart. In this paper, we present a parallel, distributed heuristic for approximating the MWCP based on dynamics principles developed and studied in various branches of mathematical biology. The proposed framework centers around a recently introduced continuous characterization of the MWCP which generalizes an earlier remarkable result by Motzkin and Straus. This allows us to formulate the MWCP (a purely combinatorial problem) in terms of a continuous quadratic programming problem. One drawback associated with this formulation, however, is the presence of "spurious" solutions, and we present characterizations of these solutions. To avoid them we introduce a new regularized continuous formulation of the MWCP inspired by previous works on the unweighted problem, and show how this approach completely solves the problem. The continuous formulation of the MWCP naturally maps onto a parallel, distributed computational network whose dynamical behavior is governed by the so-called replicator equations. These are dynamical systems introduced in evolutionary game theory and population genetics to model evolutionary processes on a macroscopic scale. We present theoretical results which guarantee that the solutions provided by our clique finding replicator network are actually the ones being sought. Extensive experiments on both randomly generated and standard benchmark graphs have been conducted, and the results obtained confirm the effectiveness of the proposed approach. Key words: Maximum weight clique, replicator equations, evolutionary game theory, dynamical systems, quadratic programming. ********************* ________________________________________________________________________ Marcello Pelillo Dipartimento di Informatica Universita' Ca' Foscari di Venezia Via Torino 155, 30172 Venezia Mestre, Italy Tel: (39) 041 2908.440 Fax: (39) 041 2908.419 E-mail: pelillo at dsi.unive.it URL: http://www.dsi.unive.it/~pelillo From movellan at markov.ucsd.edu Tue Jul 18 20:52:54 2000 From: movellan at markov.ucsd.edu (movellan) Date: Tue, 18 Jul 2000 17:52:54 -0700 Subject: Tech Reports Message-ID: <3974FBE6.AC730EFD@markov.ucsd.edu> The following technical reports are available online at http://mplab.ucsd.edu (follow links to Tech Reports) A Multithreaded Approach to Real Time Face Tracking Boris E. Shpungin and Javier R. Movellan We present a multithreaded approach to real-time face tracking. The system consists of a set of modules running on different processing threads and coordinated by a mini operating system. The first module uses color, shape, and face dynamics to determine a 2-D region of interest. The second module is a mixture of linear experts classifier whose task is to help determine whether the region of interest is indeed a face. ======================= Active Inference in Concept Induction Jonathan D. Nelson and Javier R. Movellan People are active information gatherers, constantly experimenting and seeking information relevant to their goals. A reasonable approach to active information gathering is to ask questions and conduct experiments that maximize the expected information gain given our current beliefs (Lindley, 1956; Good, 1966; MacKay, 1992). In this paper we compare the behavior of human subjects with that of an optimal information-gathering agent (infomax) in a concept induction task (Tennenbaum, 1999). Results show high consistency between subjects in their information gathering behavior. However infomax generally fails to predict what subjects do. It is unclear at this point whether the failure of infomax is due to problems with Tennenbaum's concept induction model, or due to the fact that subjects use suboptimal heuristics (e.g., confirmatory sampling). From C.Campbell at bristol.ac.uk Tue Jul 18 14:12:42 2000 From: C.Campbell at bristol.ac.uk (Colin Campbell) Date: Tue, 18 Jul 2000 19:12:42 +0100 (BST) Subject: Postdoctoral Position/Computational Intelligence Message-ID: EPSRC Postdoctoral Position available in Computational Intelligence Support Vector Machines and other Kernel-based Learning Methods Applications are invited for a EPSRC funded 2 year postdoctoral position in computational intelligence. The principal area of investigation will be kernel methods (for example, Support Vector Machines) and their applications. The emphasis of the proposed research will be the development of new algorithms (e.g. for novelty detection, classification, query learning, etc) and theoretical work to validate this approach. The project will also include some investigation of the application of kernel methods to bioinformatics, medical decision support and engineering in collaboration with research groups in these areas. The position is open to citizens of any nationality. Further details about the proposed research area may be obtained from our webpage http://lara.enm.bris.ac.uk/cig which has a downloadable review paper (An Introduction to Kernel Methods) describing the subject in more detail. The closing date for applications is *** 20th August 2000 *** Further details can be obtained from: Dr. Colin Campbell, Dept. of Engineering Mathematics, Queen's Building, University of Bristol, Bristol BS8 1TR, United Kingdom Email: C.Campbell at bris.ac.uk Tel: +44 (0) 117 928 9858 Fax: +44 (0) 117 925 1154 From mkm at hnc.com Thu Jul 20 19:48:24 2000 From: mkm at hnc.com (McClarin, Melissa) Date: Thu, 20 Jul 2000 16:48:24 -0700 Subject: HNC Financial Solutions seeks a Staff Scientist Message-ID: <72A838A51366D211B3B30008C7F4D36303FEEBDE@pchnc.hnc.com> At HNC Software, innovation and growth make us one of the most dynamic software companies in the world. We have been recognized by Forbes (ASAP Dynamic 100), Fortune (100 Fastest growing companies in America), and Software Magazine (500 Top Software Companies). Also, the San Diego Business Journal listed us as one of "The Best Companies to Work for in San Diego." HNC Financial Solutions, a division of HNC Software, is a world leader in the development and delivery of predictive, neural networks based software solutions for the financial industry. We offer the opportunity to work with cutting edge technology and a world class team in a casual atmosphere. Benefits include three weeks vacation, stock option grants and tuition reimbursement. HNC Financial Solutions is seeking a full time Staff Scientist to be based at our headquarters in San Diego, California. Please see job description and requirements below. To apply, please use one of the following methods: Mail: 5935 Cornerstone Ct. W San Diego, CA 92121 Email: hireme at hnc.com Online:http://www.hnc.com To apply for this position, please reference job code F142SD Staff Scientist Duties/Job Description: Responsibilities include designing and building predictive models based on the latest technologies in neural networks, pattern recognition/artificial intelligence, and statistical modeling for various applications in the financial industry. Specific responsibilities may vary by projects but will include analyzing data to determine suitability for modeling, pattern identifications and feature (variable) selection from large amounts of data, experimenting with different types of models, analyzing performance, and reporting results to customers. Required Qualifications (Experience/Skills): MS or PHD in Computer Science, Electrical Engineering, Applied Statistics/Math, or related field. Minimum 2 years of experience in pattern recognition, mathematical modeling, or data analysis on real world problems. Familiarity with the latest modeling techniques and tools. Good oral and written communication skills, both in terms of interacting with customers and co-workers. Very comfortable with UNIX and C. Familiar with SAS, or other analysis tools. Preferred Qualifications (Experience/Skills): Strong mathematical appetite, problem solving and computer skills. Good UNIX scripting and rapid prototyping skill. Quick learner and a good team player. Experience in designing systems based on neural networks, pattern recognition, and/or statistical modeling techniques for financial, health care, marketing, or other real world applications. From danny.silver at acadiau.ca Thu Jul 27 12:35:06 2000 From: danny.silver at acadiau.ca (Daniel L. Silver) Date: Thu, 27 Jul 2000 13:35:06 -0300 Subject: "Daniel L. Silver": PhD thesis available: Selective Transfer of Neural Network Task Knowledge Message-ID: <200007271637.e6RGbUK29773@garlic.acadiau.ca> Dear Connectionists, This is to announce the availability of my PhD thesis for download. Title: Selective Transfer of Neural Network Task Knowledge Postscript URL: http://dragon.acadiau.ca/~dsilver/DLS_THESIS.PS Zip of PS URL: http://dragon.acadiau.ca/~dsilver/DLS_THESIS.zip (323 pages) Best regards, Danny Silver ============== Keywords: task knowledge transfer, artificial neural networks, sequential learning, inductive bias, task relatedness, knowledge based inductive learning, learning to learn, knowledge consolidation Abstract: Within the context of artificial neural networks (ANN), we explore the question: How can a learning system retain and use previously learned knowledge to facilitate future learning? The research objectives are to develop a theoretical model and test a prototype system which sequentially retains ANN task knowledge and selectively uses that knowledge to bias the learning of a new task in an efficient and effective manner. A theory of {\em selective functional transfer} is presented that requires a learning algorithm that employs a {\em measure of task relatedness}. $\eta$MTL is introduced as a knowledge based inductive learning method that learns one or more secondary tasks within a back-propagation ANN as a source of inductive bias for a primary task. $\eta$MTL employs a separate learning rate, $\eta_k$, for each secondary task output $k$. $\eta_k$ varies as a function of a measure of relatedness, $R_k$, between the $k^{th}$ secondary task and the primary task of interest. Three categories of {\em a priori} measures of relatedness are developed for controlling inductive bias. The {\em task rehearsal method} (TRM) is introduced to address the issue of sequential retention and generation of learned task knowledge. The representations of successfully learned tasks are stored within a {\em domain knowledge} repository. {\em Virtual training examples} generated from domain knowledge are rehearsed as secondary tasks in parallel with each new task using either standard multiple task learning (MTL) or $\eta$MTL. TRM using $\eta$MTL is tested as a method of selective knowledge transfer and sequential learning on two synthetic domains and one medical diagnostic domain. Experiments show that the TRM provides an excellent method of retaining and generating accurate functional task knowledge. Hypotheses generated are compared statistically to single task learning and MTL hypotheses. We conclude that selective knowledge transfer with $\eta$MTL develops more effective hypotheses but not necessarily with greater efficiency. The {\em a priori} measures of relatedness demonstrate significant value on certain domains of tasks but have difficulty scaling to large numbers of tasks. Several issues identified during the research indicate the importance of consolidating a representational form of domain knowledge. ====================================================== Daniel L. Silver Danny.Silver at AcadiaU.ca Assistant Professor Intelligent Information Technology Research Centre Jodery School of Computer Science, Office 315 Acadia University ph:(902)585-1105 fax:(902)585-1067 Wolfville, NS B0P 1X0 From oby at cs.tu-berlin.de Fri Jul 28 09:18:52 2000 From: oby at cs.tu-berlin.de (Klaus Obermayer) Date: Fri, 28 Jul 2000 15:18:52 +0200 (MET DST) Subject: positions available Message-ID: <200007281318.PAA01731@pollux.cs.tu-berlin.de> Postdoctoral and Graduate Student Positions Available Neural Information Processing Group, Department of Computer Science Technical University of Berlin, Berlin, Germany The Neural Information Processing Group anticipates several openings starting this fall: 1 postdoctoral position, salary level BAT IIa 1 graduate student position, salary level, 2/3 BAT IIa Both positions are available for candidates, who are interested in developing algorithms and tools for: - 3D image processing of confocal microscope data - processing of functional imaging data (Ca-imaging, optical recording, fMRI) The candidates are expected to join a BMBF-sponsored collaboration with two computer science and four experimental neurobiology groups. We expect the postdoctoral position to be available as early as Sept. 1st 2000. The position will be initially for one year, but an extension up to three years is possible. The graduate student position is available immediately for a duration of initially two years. 2 graduate student positions, salary level BAT IIa: The positions are for candidates, who are interested in the following research areas: - computational models of visual cortex - theory of unsupervised learning algorithms Teaching duties include 4 hours tutoring per week (in German) during the winter and summer terms, i.e.\ the successful candidate must be fluent in the German language. Both positions are available Oct. 1st 2000 for a duration of five years maximum. The CS department of the Technical University of Berlin approves students with foreign Masters or Diplom degrees for graduate study when certain standards w.r.t. the subject, grades, and the awarding institution are met. The university also allows for a thesis defense in the English language. Interested candidates please send their CV, transcripts of their certificates, a short statement of their research interest, and a list of publications to: Prof. Klaus Obermayer FR2-1, NI, Informatik, Technische Universitaet Berlin Franklinstrasse 28/29, 10587 Berlin, Germany phone: ++49-30-314-73120, fax: -73121, email: oby at cs.tu-berlin.de prefereably by email. For a list of relevant publications and an overview of current research projects please refer to our web-page at: http://ni.cs.tu-berlin.de/ Cheers Klaus --------------------------------------------------------------------------- Prof. Dr. Klaus Obermayer phone: 49-30-314-73442 FR2-1, NI, Informatik 49-30-314-73120 Technische Universitaet Berlin fax: 49-30-314-73121 Franklinstrasse 28/29 e-mail: oby at cs.tu-berlin.de 10587 Berlin, Germany http://ni.cs.tu-berlin.de/ From gary at cs.ucsd.edu Fri Jul 28 13:23:30 2000 From: gary at cs.ucsd.edu (Gary Cottrell) Date: Fri, 28 Jul 2000 10:23:30 -0700 (PDT) Subject: Post doctoral positions in the PEN network Message-ID: <200007281723.KAA18891@gremlin.ucsd.edu> Postdoctoral Research Positions in Cognitive Neuroscience Applications are invited for four 2-year postdoctoral positions starting between 1/1/1 and 1/1/2 in the Perceptual Expertise Network (PEN), recently funded by the James S. McDonnell Foundation. There is no application deadline, and appropriate applicants will be chosen as they apply. The network's goals are to explore how "different" brains approach object recognition and categorization by comparing normals, agnosics, autistic individuals, non-human primates, and computational models. The network is headed by Isabel Gauthier (Vanderbilt University) and includes the following PIs: James Tanaka (Oberlin College), Tim Curran (Case Western University), Michael Tarr (Brown University), Marlene Behrmann (Carnegie Mellon University), Daniel Bub (University of Victoria), Robert Schultz (Yale Child Study Center), David Sheinberg (Brown University) and Garrison Cottrell (UCSD). One of the positions is to work at Vanderbilt with Dr. Gauthier, using fMRI and psychophysics to study the neural bases of visual object recognition. Two other positions may be held at any of the network's sites, with preference given to applicants wishing to spend time at multiple sites and demonstrating enthusiasm and skill for integrative and collaborative work. An additional 2-year position is available at the University of Victoria (under Dr. Bub's supervision), beginning 1/1/2. Applicants should have strong interests in detailed analyses of neuropsychological cases. All applicants should demonstrate an outstanding background in conducting original research on the mechanisms of face and/or object recognition, or work that has theoretical relevance in this field. For all positions, the network will support attendance to triannual PEN meetings. Post-doctoral PEN fellows will have access to a many cutting-edge techniques in cognitive neuroscience, including a computer graphics designer for stimuli preparation, fMRI, ERP, computational modeling, psychophysics, neurophysiology and neuropsychological populations (individuals with visual agnosia or autism spectrum disorders). Applicants for these positions should submit C.V., relevant reprints, 3 letters of references and a 1-2 page proposal on how they hope to combine the techniques or approaches of at least two of the PEN investigators to pursue issues in visual object representation and expertise. Applications in electronic format are encouraged (Word98 or PDF) and letters of recommendation can be sent electronically to isabel.gauthier at vanderbilt.edu. Applications can also be sent by mail to: Isabel Gauthier, Ph.D. Assistant Professor, Psychology Departmen, 502 Wilson Hall, Vanderbilt University, Nashville, TN 37240. Phone: 615 322 1778 FAX: 615 343 8449, isabel.gauthier at vanderbilt.edu. More information can be found at: http://www.psy.vanderbilt.edu/faculty/gauthier/PENpostdocs.html cheers, gary Gary Cottrell 858-534-6640 FAX: 858-534-7029 Faculty Assistant Karen Chang: 858-822-3286 kkchang at cs.ucsd.edu Computer Science and Engineering 0114 IF USING FED EX INCLUDE THE FOLLOWING LINE: "Only connect" 3101 Applied Physics and Math Building University of California San Diego -E.M. Forster La Jolla, Ca. 92093-0114 Email: gary at cs.ucsd.edu or gcottrell at ucsd.edu Home page: http://www-cse.ucsd.edu/~gary/ From: esann To: "Connectionists at cs.cmu.edu" References: From bogus@does.not.exist.com Fri Jul 28 14:30:23 2000 From: bogus@does.not.exist.com () Date: Fri, 28 Jul 2000 20:30:23 +0200 Subject: Call for sessions ESANN'2001 European Symposium on Artificial Neural Networks Message-ID: ---------------------------------------------------- | | | ESANN'2001 | | | | 8th European Symposium | | on Artificial Neural Networks | | | | Bruges (Belgium) - April 25-26-27, 2001 | | | | Call for special sessions | ---------------------------------------------------- The organizing committee of the ESANN'2001 conference is looking for proposals or suggestions concerning the organization of special sessions during the conference. Details on the organization of special sessions are available at the URL http://www.dice.ucl.ac.be/esann/callforspecialsessions.htm. More information about the ESANN'2001 conference is available at http://www.dice.ucl.ac.be/esann/. The ESANN'2001 conference will focus on fundamental aspects of ANNs: theory, models, learning algorithms, mathematical aspects, approximation of functions, classification, control, time-series prediction, statistics, signal processing, vision, self-organization, vector quantization, evolutive learning, psychological computations, biological plausibility, etc. Papers on links and comparisons between ANNs and other domains of research (such as statistics, data analysis, signal processing, biology, psychology, evolutive learning, bio-inspired systems, etc.) are encouraged. The conference will be held in Bruges (also called "Venice of the North"), one of the most beautiful medieval towns in Europe, with easy access to international travel connections. For any proposal and suggestion, please send an e-mail before August 15, 2000 to mailto:esann at dice.ucl.ac.be. ===================================================== ESANN - European Symposium on Artificial Neural Networks http://www.dice.ucl.ac.be/esann * For submissions of papers, reviews,... Michel Verleysen Univ. Cath. de Louvain - Microelectronics Laboratory 3, pl. du Levant - B-1348 Louvain-la-Neuve - Belgium tel: +32 10 47 25 51 - fax: + 32 10 47 25 98 mailto:esann at dice.ucl.ac.be * Conference secretariat D facto conference services 27 rue du Laekenveld - B-1080 Brussels - Belgium tel: + 32 2 420 37 57 - fax: + 32 2 420 02 55 mailto:esann at dice.ucl.ac.be ===================================================== From rjb at biols.susx.ac.uk Mon Jul 31 12:20:16 2000 From: rjb at biols.susx.ac.uk (Roland Baddeley) Date: Mon, 31 Jul 2000 17:20:16 +0100 Subject: Statistical analysis of visual signaling in Cuttlefish Message-ID: <3985A740.24D278EA@biols.susx.ac.uk> Three year postdoctoral position funded by the BBSRC to work on the computational analysis of visual signalling in cuttlefish, to work in collaboration with Dr Daniel Osorio and Dr Roland Baddeley at Sussex University. Cuttlefish have a remarkable ability to change their skin coloration patterns giving them amazing camouflage abilities, and a very effective visual communication channel. Previous work on cuttlefish has relied on qualitative classification of their body patterns, but by using digital video cameras, image warping techniques, and statistical pattern analysis, it is now possible to quantitatively analyze these changing skin patterns. Doing this we hope to understand their role in camouflage and communication. We are therefore looking for a postdoctoral researcher (for up to 3 years) who will: 1. Analyze the skin patterns generated by the cuttlefish. We have developed warping and independent components analysis software. The researcher would be expected implement additional means of analysis that would shed light on the dimensionality, any clustering of the patterns (expected), and implement methods to study the dynamics of the change. 2. Conduct experiments on camouflage. The animals are kept in tanks where ``textures'' can be inserted underneath. We can therefore analyze camouflage by quantifying how the cuttlefish maps the texture pattern inserted, onto its skin patterns. This will give unprecedented insight into texture and form perception by a non-human animal. As well as conducting the experiments, the researcher would be expected to maintain the animals. No previous experience is required as training will be provided. 3. Conduct experiments on communication. The cuttlefish communicate by generating stereotypical patterns on the skin. As well as analyzing these patterns on their own, given a quantitative description, the relationship between the patterns of two communicating animals can be analyzed (information capacity, relationship to behavior). Again the researcher would be expected to collect data and analyze the results. Applicants should have a strong background in statistical/neural network techniques and an interest in applying them to real world problems. The candidates will be experience in programming (MATLAB preferred) and mathematical analysis. Enquiries are welcome and can be made to Daniel Osorio or Roland Baddeley The salary will be on the standard BBSRC scale starting from 19,428 p.a. plus benefits. Applicants for this position should submit C.V., relevant reprints, and 3 letters of references. Letters of recommendation can be sent electronically (plain text, word, or PDF) to rjb at biols.susx.ac.uk Applications can also be sent by mail to: Roland Baddeley, Laboratory of Experimental Psychology, Sussex University, Falmer, Brighton England BN1 8QG. Replies will only be made to people who get called for interview. Closing date 30th October 2000. -- Dr Roland Baddeley Laboratory of Experimental Psychology Sussex University, Falmer Tel: (01273) 678961 Fax: (01273) 678611 WWW: http://www.biols.susx.ac.uk/Home/Roland_Baddeley/ From vaina at engc.bu.edu Mon Jul 31 15:59:32 2000 From: vaina at engc.bu.edu (Lucia M. Vaina) Date: Mon, 31 Jul 2000 15:59:32 -0400 Subject: T POSITION in fMRI studies of neuroplasticity--PLEASE POST!! Message-ID: Hi I would appreciate it if you could post this research position in my laboratory. Thank you. ******* The Brain and Vision Research Laboratory has an opening in the area of psychophysics and fMRI studies of adult cortical plasticity (stroke recovery and perceptual learning). An exciting new position is available for a full-time research assistant to work in Dr. Lucia Vaina's lab at Boston University http://www.bu.edu/eng/labs/bravi Our lab carries out computational and experimental research on the patterns of adult plasticity after extrastriate lesions (due to embolic stroke), perceptual learning and aspects of visual perception relevant to linking perception and action. We use psychophysics and fMRI to study both stroke patients and neurologically intact subjects. This job primarily involves assisting with fMRI experiments, including the preparation of stimulus materials, the running of subjects in an MRI scanner, the analysis of MRI data, and the preparation of this data in figures for publications and presentations. Requirements: 1. experience with multiple computer platforms ( Mac&PC, Microsoft Office, Unix, and Linux). Some knowledge of programming languages (e.g., Matlab and C) is desirable. 2. ability to trouble-shoot various hardware and software problems with the lab computers (e.g. printers, extension conflicts, network connections, etc). 3. excellent organizational and interpersonal skills. The position will involve maintaining the equipment and ordering supplies,conducting literature searches, and making appointments for subjects. 4. willingness to work on "odd hours" (depending on the magnet-time allocated to our group) and occasionally on late nights/weekends are both essential. 5. knowledge of and interest in visual cognition, neuroanatomy, and especially statistics is a great plus. This job is a good match for anyone interested in research in the neurology of vision or in cognitive neuroscience to learn more about these fields by becoming a member of a lively and productive interdisciplinary research group housed at multiple sites in Boston. *** Salary commensurable with experience*** The job can begin any time after September 1. Commitment of one year is required. Send inquiries, CV and names and phone numbers of 2-3 references to vaina at bu.edu or best to Professor Lucia M. Vaina Lucia M. Vaina Ph.D., Sc.D. Brain and Vision Research Laboratory Boston University, Department of Biomedical Engineering College of Engineering 44 Cummington str, Room 315 Boston University Boston, Ma 02215 USA tel: 617-353-2455 fax: 617-353-6766 From sshams at biodiscovery.com Mon Jul 31 12:14:31 2000 From: sshams at biodiscovery.com (Soheil Shams) Date: Mon, 31 Jul 2000 09:14:31 -0700 Subject: Job Posting: Senior Image Processing Scientist Message-ID: Senior Image Processing Scientist Description: We are looking for a talented, creative individual with a strong background in image processing, machine vision, and neural networks to direct and participate in the development of existing and new image processing products for genetic analysis. This position involves development and implementation of machine vision and image processing algorithms, such as image segmentation, multi-spectral image processing, and statistical and ANN based image quality assessment, customer interactions, research, and providing educational training services. The position requires the ability to formulate problem descriptions through interaction with end-user customers. These technical issues must then be transformed into innovative and practical algorithmic solutions. We expect our scientists to have outstanding written/oral communication skills and encourage publications and presentations in scientific journals and conferences. Requirements: - A Ph.D. in Electrical Engineering, Computer Science, or related field, or equivalent experience. - Experience developing image processing and pattern recognition algorithms. - Solid experience with Object Oriented programming in Java and/or C++ - Knowledge of biology and genetics is a plus but not necessary. - Excellent communication skills. For consideration, please send your r?sum? along with a cover letter to E-mail: HR at BioDiscovery.com Fax: (310) 966-9346 Snail-mail: BioDiscovery, Inc. 11150 W. Olympic Blvd. Suite 1170 Los Angeles, CA 90064 BioDiscovery, Inc. is start-up company founded in 1997 and dedicated to the development of state-of-the-art bioinformatics tools for biotechnology research applications. We are a leading gene expression image and data analysis firm with an outstanding client list and a progressive industry stance. BioDiscovery is an equal opportunity employer with good benefits, and our shop has a friendly, high-energy atmosphere. We are headquartered in sunny Southern California near the UCLA campus. From nschweighofer at neurotek.co.jp Mon Jul 31 02:37:09 2000 From: nschweighofer at neurotek.co.jp (Nicolas Schweighofer) Date: Mon, 31 Jul 2000 15:37:09 +0900 Subject: job offer in Tokyo Message-ID: <592E79EA9603D411B08F00D0B7203CB1053EC0@LCTKY01> My company has two research positions that could be of interest for people with training in statistics/neural networks. Learning Curve K.K, Tokyo, Japan: We have developed a new, scientific, and patented learning method, which has the potential to radically change the way we learn. Our company is now preparing to introduce this method to both the Japan and U.S. markets in the form of computer software applications, and we are expanding our research team for the development and testing of future versions of this method. The positions require a Masters or a PhD degree with training in computer science, applied math, physics, psychology, cognitive neuroscience, or related field. Depending on the position, the candidates should either have strong analytical skills preferably with experience in statistics/neural networks or have proven experience in designing and conducting experiments with subjects. Applicants should be creative, have the ability to quickly learn new techniques and concepts, and have excellent oral and written skills in either English or Japanese. We offer very competitive salaries and an employee Stock Option Plan. If you are looking for the opportunity to conduct high-level research in an exciting pre-IPO start-up in Tokyo, please send your resume to nicolas at neurotek.co.jp. Nicolas Schweighofer, Ph.D. Director of Core Research The Learning Curve, Inc. Tel: +81-3-3463-7266 Fax: +81-3-5489-7015 nicolas at neurotek.co.jp