REVIEW Communicated by Veronka Cheplygina Indefinite Proximity Learning: A Review Frank-Michael Schleif schleif@cs.bham.ac.uk Peter Tino P.Tino@bham.ac.uk University of Birmingham, School of Computer Science, B15 2TT, Birmingham, REINO UNIDO. Efficient learning of a data analysis task strongly depends on the data representation. Most methods rely on (symmetric) similarity or dissim- ilarity representations by means of metric inner products or distances, providing easy access to powerful mathematical formalisms

LETTER Communicated by Harel Shouval Two-Trace Model for Spike-Timing-Dependent Synaptic Plasticity Rodrigo Echeveste echeveste@itp.uni-frankfurt.de Claudius Gros gros07@itp.uni-frankfurt.de Institute for Theoretical Physics, Goethe University Frankfurt, Hessen 60438, Germany We present an effective model for timing-dependent synaptic plasticity (STDP) in terms of two interacting traces, corresponding to the fraction of activated NMDA receptors and the Ca2+ concentration in the den- dritic spine of the postsynaptic neuron. Este

ARTICLE Communicated by Minoru Asada A Neural Framework for Organization and Flexible Utilization of Episodic Memory in Cumulatively Learning Baby Humanoids Vishwanathan Mohan vishwanathan.mohan@iit.it Giulio Sandini giulio.sandini@iit.it Pietro Morasso pietro.morasso@iit.it Robotics, Brain and Cognitive Science Department, Istituto Italiano di Tecnologia, Genova, Italy Cumulatively developing robots offer a unique opportunity to reenact the constant interplay between neural mechanisms related to learning, memory, prospection, and abstraction from

ARTICLE Communicated by Adam Kampff High-Dimensional Cluster Analysis with the Masked EM Algorithm Shabnam N. Kadir s.kadir@ucl.ac.uk UCL Institute of Neurology and UCL Department of Neuroscience, Physiology, and Pharmacology, University College London, London WC1E 6DE, REINO UNIDO. Dan F. METRO. Goodman Dan_Goodman@meei.harvard.edu Eaton-Peabody Laboratories, Massachusetts Eye and Ear Infirmary, Department of Otology and Laryngology, Harvard Medical School, Bostón, MAMÁ 02114, U.S.A. Kenneth D. Harris kenneth.harris@ucl.ac.uk UCL

Communicated by Richard Andersen Computing Optical Flow in the Primate Visual System H. Taichi Wang Bimal Mathur Science Center, Rockwell International, Thousand Oaks, California 91360, USA Christof Koch Computation and Neural Systems Program, Divisions of Biology and Engineering and Applied Sciences, 21 6-76, California Institute of Technology, Pasadena, California 91 125, USA Computing motion on the basis of the time-varying image intensity is a difficult

Communicated by Dana Ballad Part Segmentation for Object Recognition Alex Pentland Vision Sciences Group, The Media Lab, Instituto de Tecnología de Massachusetts, Room E15-410, 20 Ames Street, Cambridge, M A 02139, USA Visual object recognition is a difficult problem that has been solved by biological visual systems. An approach to object recognition is described in which the image is segmented into parts using two simple, biologically-plausible

Communicated by David Mumford Two Stages of Curve Detection Suggest Two Styles of Visual Computation Steven W. Zuckert AIlan Dobbins Lee Iverson Computer Vision and Robotics Laboratory, McGill Research Centre for Intelligent Machines, Universidad McGill, Montrkal, QuCbec, Canada The problem of detecting curves in visual images arises in both com- puter vision and biological visual systems. Our approach integrates constraints from these two sources and

Communicated by Carver Mead Criteria for Robust Stability In A Class Of Lateral Inhibition Networks Coupled Through Resistive Grids John L. Wyatt, Jr. David L. Standley Department of EJectrical Engineering and Computer Science, Massachusetts Institute of TechnoJogy, Cambridge, MAMÁ 02139, USA In the analog VLSI implementation of neural systems, it is sometimes convenient to build lateral inhibition networks by using a locally con- nected on-chip

Communicated by John Wyatt A Silicon Model Of Auditory Localization John Lazzaro Carver A. Mead Department of Computer Science, California Institute of Technology, MS 256-80, Pasadena, California 91125, USA The barn owl accurately localizes sounds in the azimuthal plane, a nosotros- ing interaural time difference as a cue. The time-coding pathway in the owl’s brainstem encodes a neural map of azimuth, by processing interaural timing information.

Communicated by Richard Lippmann Modular Construction of Time-Delay Neural Networks for Speech Recognition Alex Waibel Computer Science Department, Carnegie Mellon University, pittsburgh, Pensilvania 15213, USA and ATR Interpreting Telephony R.esearch Laboratories, Twin 21 MiD Tower, Osaka, 540, Japan Several strategies are described that overcome limitations of basic net- work models as steps towards the design of large connectionist speech recognition systems. The two major areas

Communicated by Les Valiant . What Size Net Gives Valid Generalization? Eric B. Baum* Jet Propulsion Laboratorj; California Institute o f Technology, Pasadena, C A 91 109, USA David Haussler Department o f Computer and Information Sciencrs, University o f California, Santa Cruz, California 95064, USA We address the question of when a network can be expected to general- ize from m random training examples

Communicated by Scott Kirkpatrick Deterministic Boltzmann Learning Performs Steepest Descent in Weight-Space Geoffrey E. Hinton Department of Computer Science, University o f Toronto, 10 King’s College Road, Toronto M5S 1 A4, Canada The Boltzmann machine learning procedure has been successfully ap- plied in deterministic networks of analog units that use a mean field ap- proximation to efficiently simulate a truly stochastic system (Peterson and Anderson

Communicated by David Touretzky Product Units: A Computationally Powerful and Biologically Plausible Extension to Backpropagation Networks Richard Durbin David E. Rumelhart Department of Psycholoa, Universidad Stanford, stanford, California 94305, USA We introduce a new form of computational unit for feedfoxward learn- ing networks of the backpropagation type. Instead of calculating a weighted sum this unit calculates a weighted product, where each in- put is raised

Communicated by Patricia Churchland The Brain Binds Entities and Events by Multiregional Activation from Convergence Zones Antonio R. Damasio Department of Neurology, Division of Behavioral Neurology and Cognitive Neuroscience, University of Iowa College of Medicine, Iowa City, IA, USA The experience of reality, in both perception and recall, is spatially and temporally coherent and «in-register.» Features are bound in enti- corbatas, and entities are bound

Review of Neural Networks for Speech Recognition Richard P. Lippmann* MIT Lincoln Laboratory, Lexington, MAMÁ 021 73, USA The performance of current speech recogition systems is far below that of humans. Neural nets offer the potential of providing massive parallelism, adaptación, and new algorithmic approaches to problems in speech recognition. Initial studies have demonstrated that multi- layer networks with time delays can provide excellent discrimination

Communicated by David A. Robinson A Control Systems Model of Smooth Pursuit Eye Movements with Realistic Emergent Properties R. j. Krauzlis S. GRAMO. Lisberger Department of Physiology and Neuroscience Graduate Program, Universidad de California, San Fkancisco, C A 94143, USA Visual tracking of objects in a noisy environment is a difficult problem that has been solved by the primate oculomotor system, but remains unsolved in

Communicated by Richard Lippmann A Multiple-Map Model for Pattern Classification Alan Rojer Eric Schwartz Computational Neuroscience Laboratory, New York University Medical Center, Courant Institute of Mathematical Sciences, New York University, Nueva York, Nueva York 1001 6, USA A characteristic feature of vertebrate sensory cortex (and midbrain) is the existence of multiple two-dimensional map representations. Some workers have considered single-map classification (p.ej. Kohonen 1984) but little work