ADVANCES IN NEURAL INFORMATION 
PROCESSING SYSTEMS 11 
ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 
Published by Morgan-Kaufmann 
NIPS-1 
Advances in Neural Information Processing Systems 1: Proceedings of the 1988 Conference, 
David S. Touretzky, ed., 1989. 
NIPS-2 
Advances in Neural Information Processing Systems 2: Proceedings of the 1989 Conference, 
David S. Touretzky, ed., 1990. 
NIPS-3 
Advances in Neural Information Processing Systems 3.' Proceedings of the 1990 Conference, 
Richard Lippmann, John E. Moody and David S. Touretzky, eds., 1991. 
NIPS-4 
Advances in Neural Information Processing Systems 4: Proceedings of the 1991 Conference, 
John E. Moody, Stephen J. Hanson and Richard P. Lippmann, eds., 1992. 
NIPS-5 
Advances in Neural Information Processing Systems 5.' Proceedings of the 1992 Conference, 
Stephen J. Hanson, Jack D. Cowan and C. Lee Giles, eds., 1993. 
NIPS-6 
Advances in Neural Information Processing Systems 6: Proceedings of the 1993 Conference, 
Jack D. Cowan, Gerald Tesauro and Joshua Alspector, eds., 1994. 
Published by The MIT Press 
NIPS-7 
Advances in Neural Information Processing Systems 7.' Proceedings of the 1994 Conference, 
Gerald Tesauro, David S. Touretzky and Todd K. Leen, eds., 1995. 
NIPS-8 
Advances in Neural Information Processing Systems 8: Proceedings of the 1995 Conference, 
David S. Touretzky, Michael C. Mozer and Michael E. Hasselmo, eds., 1996. 
NIPS-9 
Advances in Neural Information Processing Systems 9: Proceedings of the 1996 Conference, 
Michael C. Mozer, Michael I. Jordan and Thomas Petsche, eds., 1997. 
NIPS-10 
Advances in Neural Information Processing Systems 10: Proceedings of the 1997 Conference, 
Michael I. Jordan, Michael J. Keams and Sara A. Solla, eds., 1998. 
NIPS-11 
Advances in Neural Information Processing Systems 11: Proceedings of the 1998 Conference, 
Michael S. Keams, Sara A. Solla and David A. Cohn, eds., 1999. 
ADVANCES IN NEURAL INFORMATION 
PROCESSING SYSTEMS 11 
Proceedings of the 1998 Conference 
edited by 
Michael S. Kearns, Sara ,4. Solla and David,4. Cohn 
A Bradford Book 
The MIT Press 
Cambridge, Massachusetts 
London, England 
 1999 Massachusetts Institute of Technology 
All rights reserved. No part of this book may be reproduced in any form by any electronic or 
mechanical means (including photocopying, recording or information storage and retrieval) 
without permission in writing from the publisher. 
This book was printed and bound in the United States of America. 
ISSN: 1049-5258 
ISBN: 0-262-11245-0 
Contents 
Preface 
NIPS Committees 
Reviewers 
xv 
xvii 
xix 
Part II Neuroscience 
Temporally Asymmetric Hebbian Learning, Spike liming and Neural Response 
Variability, L. F. Abbott and Sen Song ...................... 69 
Contrast Adaptation in Simple Cells by Changing the Transmitter Release 
Probability, Peter Adorjtin and Klaus Obermayer ................. 76 
Where Does the Population Vector of Motor Cortical Cells Point during 
Reaching Movements?, Pierre Baraduc, Emmanuel Guigon and Yves Burnod . . 83 
Recurrent Cortical Amplification Produces Complex Cell Responses, 
Frances S. Chance, Sacha B. Nelson and L. F. Abbott ............... 90 
Neuronal Regulation Implements Efficient Synaptic Pruning, 
Gal Chechik, Isaac Meilijson and Eytan Ruppin ................. 97 
Divisive Normalization, Line Attractor Networks and Ideal Observers, 
Sophie Deneve, Alexandre Pouget and Peter E. Latham .............. 104 
Part I Cognitive Science 
Evidence for a Forward Dynamics Model in Human Adaptive Motor Control 
Nikhil Bhushan and Reza Shadmehr ....................... 3 
Perceiving without Learning.' From Spirals to Inside/Outside Relations, 
Ke Chen and DeLiang L. Wang .......................... 10 
A Model for Associative Multiplication, 
G. Bj6rn Christianson and Suzanna Becker .................... 17 
Facial Memory Is Kernel Density Estimation (Almost), 
Matthew N. Dailey, Garrison W. Cottrell and Thomas A. Busey .......... 24 
Multiple Paired Forward-Inverse Models for Human Motor Learning and 
Control, Masahiko Hamno, Daniel M. Wolpert and Mitsuo Kawato ........ 31 
Utilizing lime: Asynchronous Binding, Bradley C. Love ............. 38 
Mechanisms of Generalization in Perceptual Learning, 
Zili Liu and Daphna Weinshall .......................... 45 
A Principle for Unsupervised Hierarchical Decomposition of Visual Scenes, 
Michael C. Mozer ................................ 52 
Bayesian Modeling of Human Concept Learning, Joshua B. Tenenbaum ..... 59 
vi Contents 
Synergy and Redundancy among Brain Cells of Behaving Monkeys, 
Itay Gat and Naftali Tishby ............................ 111 
Analyzing and Visualizing Single-Trial Event-Related Potentials, 
Tzyy-Ping Jung, Scott Makeig, Marissa Westerfield, Jeanne Townsend, 
Eric Courchesne and Terrence J. Sejnowski .................... 
118 
Spike-Based Compared to Rate-Based Hebbian Learning, 
Richard Kempter, Wulfram Gerstner and J. Leo van Hemmen ........... 
125 
Signal Detection in Noisy Weakly-Active Dendrites, 
Amit Manwani and Christof Koch ........................ 
132 
The Role of Lateral Cortical Competition in Ocular Dominance Development, 
Christian Piepenbrock and Klaus Obermayer ................... 
139 
Multi-Electrode Spike Sorting by Clustering Transfer Functions, 
Dmitry Rinberg, Hanan Davidowitz and Naftali Tishby .............. 146 
Modeling Surround Suppression in V1 Neurons with a Statistically Derived 
Normalization Model, Eero P. Simoncelli and Odelia Schwartz .......... 
153 
Information Maximization in Single Neurons, 
Martin Stemmler and Christof Koch ....................... 
160 
The Effect of Correlations on the Fisher Information of Population Codes, 
Hyoungsoo Yoon and Haim Sompolinsky ..................... 167 
Distributional Population Codes and Multiple Motion Models, 
174 
Richard S. Zemel and Peter Dayan ........................ 
Part III Theory 
Tractable Variational Structures for Approximating Graphical Models, 
David Barber and Wim Wiegerinck ........................ 
183 
Almost Linear VC Dimension Bounds for Piecewise Polynomial Networks, 
Peter L. Bartlett, Vitaly Maiorov and Ron Meir ................ 
190 
Dynamics of Supervised Learning with Restricted Training Sets, 
A. C. C. Coolen and David Saad ......................... 
197 
Dynamically Adapting Kernels in Support Vector Machines, 
Nello Cristianini, Colin Campbell and John Shawe-Taylor ............ 204 
Phase Diagram and Storage Capacity of Sequence-Storing Neural Networks, 
A. Daring, A. C. C. Coolen and D. Sherrington .................. 211 
Finite-Dimensional Approximation of Gaussian Processes, 
Giancarlo Ferrari-Trecate, Christopher K. I. Williams and Manfred Opper ..... 218 
Linear Hinge Loss and Average Margin, 
Claudio Gentile and Manfred K. Warrnuth .................... 225 
Unsupervised and Supervised Clustering.' The Mutual Information between 
Parameters and Observations, Didier Herschkowitz and Jean-Pierre Nadal .... 
232 
Contents vii 
Convergence of the Wake-Sleep Algorithm, 
Shiro Ikeda, Shun-ichi Amari and Hiroyuki Nakahara ............... 239 
The Belief in TAP, Yoshiyuki Kabashima and David Saad ............. 246 
Optimizing Classifers for Imbalanced Training Sets, 
Grigoris Karakoulas and John Shawe-Taylor ................... 253 
Inference in Multilayer Networks via Large Deviation Bounds, 
Michael Kearns and Lawrence Saul ........................ 260 
Stationarity and Stability of Autoregressive Neural Network Processes, 
Friedrich Leisch, Adrian Trapletti and Kurt Hornik ................ 267 
Computational Differences between Asymmetrical and Symmetrical Networks, 
Zhaoping Li and Peter Dayan ........................... 274 
A Precise Characterization of the Class of Languages Recognized by Neural 
Nets under Gaussian and Other Common Noise Distributions 
Wolfgang Maass and Eduardo D. Sontag ..................... 281 
Direct Optimization of Margins Improves Generalization in Combined 
Classifiers, Llew Mason, Peter L. Bartlett and Jonathan Baxter .......... 288 
On the Optimality of Incremental Neural Network Algorithms, 
Ron Meir and Vitaly Maiorov .......................... 295 
General Bounds on Bayes Errors for Regression with Gaussian Processes, 
Manfred Opper and Francesco Vivarelli ..................... 302 
Mean FieM Methods for Classification with Gaussian Processes, 
Manfred Opper and Ole Winther ......................... 309 
On-Line Learning with Restricted Training Sets: Exact Solution as Benchmark 
for General Theories, H. C. Rae, Peter Sollich and A. C. C. Coolen ........ 316 
Ight Bounds for the VC-Dimension of Piecewise Polynomial_Networks, 
Akito Sakurai .................................. 323 
Shrinking the Tube: A New Support Vector Regression Algorithm, 
Bernhard SchOlkopf, Peter L. Bartlett, Alex J. Smola and Robert Williamson 330 
Discontinuous Recall Transitions Induced by Competition Between Short- and 
Long-Range Interactions in Recurrent Networks, 
N. S. Skantzos, C. F. Beckmann and A. C. C. Coolen ............... 337 
Learning Curves for Gaussian Processes, Peter Sollich .............. 344 
A Theory of Mean Field Approximation, Toshiyuki Tanaka ............ 351 
Part IV Algorithms and Architecture 
Learning a Hierarchical Belief Network of Independent Factor Analyzers, 
Hagai Attias ................................... 
Semi-Supervised Support Vector Machines, Kristin Bennett and Ayhan Demiriz 
361 
368 
viii Contents 
Lazy Learning Meets the Recursive Least Squares Algorithm, 
Mauro Birattari, Gianluca Bontempi and Hugues Bersini ............. 
375 
Bayesian PCA, Christopher M. Bishop ...................... 382 
Learning Multi-Class Dynamics, Andrew Blake, Ben North and Michael Isard 
389 
Approximate Learning of Dynamic Models, Xavier Boyen and Daphne Koller 
396 
Fisher Scoring and a Mixture of Modes Approach for Approximate Inference and 
Learning in Nonlinear State Space Models, Thomas Briegel and Volker Tresp 
403 
Global Optimisation of Neural Network Models via Sequential Sampling 
Joo F. G. de Freitas, Mahesan Niranjan, Amaud Doucet and Andrew H. Gee 
410 
Efficient Bayesian Parameter Estimation in Large Discrete Domains, 
Nir Friedman and Yoram Singer ......................... 417 
A Randomized Algorithm for Pairwise Clustering 
Yoram Gdalyahu, Daphna Weinshall and Michael Werrnan ............ 424 
Learning Nonlinear Dynamical Systems Using an EM Algorithm, 
Zoubin Ghahramani and Sam T. Roweis ..................... 
431 
Classification on Pairwise Proximity Data, 
Thore Graepel, Ralf Herbrich, Peter Bollmann-Sdorra and Klaus Obermayer 
438 
Outcomes of the Equivalence of Adaptive Ridge with Least Absolute Shrinkage, 
Yves Grandvalet and St6phane Canu ....................... 445 
Visualizing Group Structure, 
Marcus Held, Jan Puzicha and Joachim M. Buhmann ............... 452 
Source Separation as a By-Product of Regularization, 
Sepp Hochreiter and Jtirgen Schmidhuber .................. 
459 
Learning from Dyadic Data, 
Thomas Holmann, Jan Puzicha and Michael I. Jordan ............... 466 
Sparse Code Shrinkage.' Denoising by Nonlinear Maximum Likelihood 
Estimation, Aapo Hyvfirinen, Patrik Hoyer and Erkki Oja ............. 473 
Restructuring Sparse High Dimensional Data for Effective Retrieval, 
Charles Lee Isbell, Jr. and Paul Viola ....................... 480 
Exploiting Generative Models in Discriminative Classifiers, 
Tommi S. Jaakkola and David Haussler ...................... 487 
Maximum Conditional Likelihood via Bound Maximization and the CEM 
Algorithm, Tony Jebara and Alex Pentland .................... 494 
A PolygonalLine Algorithm for Constructing Principal Curves, 
Balfizs K6gl, Adam Krzyiak, Tamils Linder and Kenneth Zeger .......... 501 
Unsupervised Classification with Non-Gaussian Mixture Models Using ICA, 
Te-Won Lee, Michael S. Lewicki and Terrence J. Sejnowski ........... 508 
Learning a Continuous Hidden Variable Model for Binary Data, 
Daniel D. Lee and Haim Sompolinsky ...................... 515 
Contents ix 
Neural Networks for Density Estimation, Malik Magdon-Ismail and Amir Atiya 522 
Exploratory Data Analysis Using Radial Basis Function Latent Variable Models, 
Alan D. Marrs and Andrew R. Webb ....................... 529 
Kernel PCA and De-Noising in Feature Spaces, Sebastian Mika, 
Bernhard Sch61kopf, Alex J. Smola, Klaus-Robert MOller, Matthias Scholz and 
Gunnar R/tsch .................................. 536 
Very Fast EM-Based Mixture Model Clustering Using Multiresolution Kd-Trees, 
Andrew W. Moore ................................ 543 
Replicator Equations, Maximal Cliques, and Graph Isomorphism, 
Marcello Pelillo ................................. 550 
Using Analytic QP and Sparseness to Speed Training of Support Vector 
Machines, John C. Platt ............................. 557 
RegularizingAdaBoost, Gunnar R/tsch, Takashi Onoda and Klaus-Robert MOller 564 
Boxlets.' A Fast Convolution Algorithm for Signal Processing and Neural 
Networks, Patrice Y. Simard, L6on Bottou, Patrick Haffner and Yann Le Cun 571 
Batch and On-Line Parameter Estimation of Gaussian Mixtures Based on the 
Joint Entropy, Yoram Singer and Manfred K. Warmuth .............. 578 
Semiparametric Support Vector and Linear Programming Machines, 
Alex J. Smola, Thilo T. Frie13 and Bernhard Sch61kopf .............. 585 
Probabilistic Isualisation of High-Dimensional Binary Data, Michael E. Tipping 592 
SMEM Algorithm for Mixture Models, 
Naonori Ueda, Ryohei Nakano, Zoubin Ghahramani and Geoffrey E. Hinton 599 
Learning Mixture Hierarchies, Nuno Vasconcelos and Andrew Lippman ..... 606 
Discovering Hidden Features with Gaussian Processes Regression, 
Francesco Vivarelli and Christopher K. I. Williams ................ 613 
The Bias-Variance Tradeoff and the Randomized GACV, Grace Wahba, 
Xiwu Lin, Fangyu Gao, Dong Xiang, Ronald Klein and Barbara Klein ...... 620 
Basis Selection for Wavelet Regression, Kevin R. Wheeler and Atam P. Dhawan . . 627 
DTs: Dynamic Trees, Christopher K. I. Williams and Nicholas J. Adams ..... 634 
Convergence Rates of Algorithms for qsual Search: Detecting qsual Contours, 
A. L. Yuille and James M. Coughlan ....................... 641 
Blind Separation of Filtered Sources Using State-Space Approach, 
Liqing Zhang and Andrzej Cichocki ....................... 648 
Part V Implementation 
Analog VLSI Cellular Implementation of the Boundary Contour System, 
Gert Cauwenberghs and James Waskiewicz .................... 
657 
x Contents 
Active Noise Canceling Using Analog Neuro-Chip with On-Chip Learning 
Capability, Jung-Wook Cho and Soo-Young Lee ................. 664 
A Micropower CMOSAdaptive Amplitude and Shift Invariant Vector Quantiser, 
Richard J. Coggins, Raymond J. W. Wang and Marwan A. Jabri .......... 671 
Optimizing Correlation Algorithms for Hardware-Based Transient Classification, 
R. Timothy Edwards, Gert Cauwenberghs and Fernando J. Pineda ........ 678 
VLSI Implementation of Motion Centroid Localization for Autonomous 
Navigation, 
Ralph Etienne-Cummings, Viktor Gruev and Mohammed Abdel Ghani ...... 685 
A Neuromorphic Monaural Sound Localizer, 
John G. Harris, Chiang-Jung Pu and Jose C. Principe ............... 692 
An Integrated Vision Sensor for the Computation of Optical Flow Singular 
Points, Charles M. Higgins and Christof Koch .................. 699 
Computation of Smooth Optical Flow in a Feedback Connected Analog Network, 
Alan Stocker and Rodney Douglas ....................... 706 
,4 High Performance k-NN Classifier Using a Binary Correlation Matrix 
Memory, Ping Zhou, Jim Austin and John Kennedy ................ 713 
Part VI Speech, Handwriting and Signal Processing 
An Entropic Estimator for Structure Discovery, Matthew Brand .......... 723 
Coding lime- Varying Signals Using Sparse, Shift-Invariant Representations, 
Michael S. Lewicki and Terrence J. Sejnowski .................. 
Controlling the Complexity of HMM Systems by Regularization, 
Christoph Neukirchen and Gerhard Rigoil .................... 
730 
737 
Maximum-Likelihood Continuity Mapping (MALCOM) : An Alternative to 
HMMs, David A. Nix and John E. Hogden .................... 744 
Markov Processes on Curves for Automatic Speech Recognition, 
Lawrence Saul and Mazin Rahim ......................... 751 
Part VII Visual Processing 
,4 Phase Space Approach to Minimax Entropy Learning and the Minutemax 
Approximations, James M. Coughlan and A. L. Yuille ............... 761 
Example-Based Image Synthesis of Articulated Figures, Trevor Darrell ...... 768 
Learning to Estimate Scenes from Images, 
William T. Freeman and Egon C. Pasztor ..................... 775 
Learning to Find Pictures of People, Sergey Ioffe and David Forsyth ....... 782 
Attentional Modulation of Human Pattern Discrimination Psychophysics 
Reproduced by a Quantitative Model, 
Laurent Itti, Jochen Braun, Dale K. Lee and Christof Koch ............ 789 
Contents xi 
A V1 Model of Pop Out and Asymmetry in sual Search, Zhaoping Li ...... 796 
Support Vector Machines Applied to Face Recognition, P. Jonathon Phillips .... 803 
Learning Lie Groups for Invariant sual Perception, 
Rajesh P. N. Rao and Daniel L. Ruderman .................... 810 
General-Purpose Localization of Textured Image Regions, Ruth Rosenholtz .... 817 
Probabilistic Image Sensor Fusion, 
Ravi K. Sharma, Todd K. Leen and Misha Pavel ................. 824 
Orientation, Scale, and Discontinuity as Emergent Properties of Illusory Contour 
Shape, Karvel K. Thomber and Lance R. Williams ................ 831 
Classification in Non-Metric Spaces, 
Daphna Weinshall, David W. Jacobs and Yoram Gdalyahu ............ 838 
Part VIII Applications 
Mala'ng Templates Rotationally Invariant.' An Application to Rotated Digit 
Recognition, Shumeet Baluja ........................... 847 
Probabilistic Modeling for Face Orientation Discrimination: Learning from 
Labeled and Unlabeled Data, Shumeet Baluja .................. 854 
Adding Constrained Discontinuities to Gaussian Process Models of Wind Fields, 
Dan Cornford, Ian T. Nabney and Christopher K. I. Williams ........... 861 
Vertex Identification in High Energy Physics Experiments, 
Gideon Dror, Halina Abramowicz and David Horn ................ 868 
Familiarity Discrimination of Radar Pulses, 
Eric Granger, Stephen Grossberg, Mark A. Rubin and William W. Streilein .... 875 
Fast Neural Network Emulation of Dynamical Systems for Computer Animation, 
Radek Grzeszczuk, Demetri Terzopoulos and Geoffrey E. Hinton ......... 882 
Call-Based Fraud Detection in Mobile Communication Networks Using a 
Hierarchical Regime-Switching Model, Jaakko Hollm6n and Volker Tresp ..... 889 
Graph Matching for Shape Retrieval, 
Benoit Huet, Andrew D. J. Cross and Edwin R. Hancock ............. 896 
Scheduling Straight-Line Code Using Reinforcement Learning and Rollouts, 
Amy McGovern and Eliot Moss ......................... 903 
Bayesian Modeling of Facial Similarity, 
Baback Moghaddam, Tony Jebara and Alex Pentland ............... 910 
Reinforcement Learning for Trading, John Moody and Matthew SalTell ...... 917 
Graphical Models for Recognizing Human Interactions, 
Nuria M. Oliver, Barbara Rosario and Alex Pentland ............... 924 
Independent Component Analysis of lntracellular Calcium Spike Data, 
Klaus Prank, Julia B6rger, Alexander von zur Miihlen, Georg Brabant and 
Christof Sch6fl .................................. 931 
xii Contents 
Applications of Multi-Resolution Neural Networks to Mammography, 
Clay D. Spence and Paul Sajda .......................... 938 
Robot Docking Using Mixtures of Gaussians, 
Matthew M. Williamson, Roderick Murray-Smith and Volker Hansen ....... 945 
Using Collective Intelligence to Route Internet Traffic, 
David H. Wolpert, Kagan Tumer and Jeremy Frank ................ 
952 
Part IX Control, Navigation and Planning 
Robust, Efficient, Globally-Optimized Reinforcement Learning with the 
Parti-Game Algorithm, Mohammad A. A1-Ansari and Ronald J. Williams ..... 961 
Gradient Descent for General Reinforcement Learning, 
Leemon Baird and Andrew W. Moore ...................... 968 
Non-Linear PI Control Inspired by Biological Control Systems, 
Lyndon J. Brown, Gregory E. Gonye and James S. Schwaber ........... 975 
Optimizing Admission Control while Ensuring Quality of Service in Multimedia 
Networks via Reinforcement Learning, 
Timothy X. Brown, Hui Tong and Satinder Singh ................. 982 
Iewing Classifier Systems as Model Free Learning in POMDPs, 
Akira Hayashi and Nobuo Suematsu ..................... 989 
Finite-Sample Convergence Rates for Q-Learning and Indirect Algorithms, 
Michael Kearns and Satinder Singh ........................ 996 
Exploring Unknown Environments with Real-71me Search or Reinforcement 
Learning, Sven Koenig .............................. 1003 
The Effect of Eligibility Traces on Finding Optimal Memoryless Policies in 
Partially Observable Markov Decision Processes, John Loch ........... 1010 
Learning Instance-Independent Value Functions to Enhance Local Search, 
Robert Moll, Andrew G. Barto, Theodore J. Perkins and Richard S. Sutton .... 1017 
Barycentric Interpolators for Continuous Space and 71me Reinforcement 
Learning, R6mi Munos and Andrew W. Moore .................. 1024 
Risk Sensitive Reinforcement Learning, Ralph Neuneier and Oliver Mihatsch 1031 
Coordinate Transformation Learning of Hand Position Feedback Controller by 
Using Change of Position Error Norm, Eimei Oyama and Susumu Tachi ..... 1038 
Learning Macro-Actions in Reinforcement Learning, Jette Randlov ........ 1045 
Reinforcement Learning Based on On-Line EM Algorithm, 
Masa-aki Sato and Shin Ishii ........................... 1052 
A Reinforcement Learning Algorithm in Partially Observable Environments 
Using Short-Term Memory, Nobuo Suematsu and Akira Hayashi ......... 1059 
Improved Switching among Temporally Abstract Actions, 
Richard S. Sutton, Satinder Singh, Doina Precup and Balaraman Ravindran .... 1066 
Contents xiii 
Experimental Results on Learning Stochastic Memoryless Policies for Partially 
Observable Markov Decision Processes, John K. Williams and Satinder Singh 
Index of Authors ................................. 
Keyword Index .................................. 
1073 
1081 
1085 
Preface 
This volume comains the papers presemed at the Twelfth Annual Conference on Neural 
Information Processing Systems (NIPS), held in Denver, Colorado from November 30 to 
December 5, 1998. These papers cover a wide range of topics in neural computation, 
including the design and analysis of learning algorithms, learning theory, reinforcement 
learning, neuroscience, cognitive science, vision, speech, VLSI implememations, and ap- 
plications in areas such as artificial intelligence, applied statistics, pattern recognition, sig- 
nal processing, control, finance, and data analysis. 
Paper selection was competitive and based on the reports of three specific referees per pa- 
per, and the global recommendations of the members of the Program Committee. This year 
we evaluated 474 submissions to select the 150 papers included in this volume. The NIPS 
acceptance ratio cominues to be lower than that of many joumals; many good papers are 
left out due to limitations imposed by the length of the Conference and the size of the Pro- 
ceedings. In a clear indication of the imellectual health of the conference and the expected 
high quality of the accepted papers, attendance remains high at about 500 participams. In 
addition to loyal followers, we are glad to welcome new faces every year, as we rely on the 
infusion of new ideas and new approaches to maimain the sense of excitemere that is the 
hallmark of NIPS. It is in this spirit that the NIPS Foundation provided financial support to 
allow many graduate studeros and young investigators to attend the meeting. 
NIPS brings together top sciemists with broadly varying backgrounds, such as statistics, 
mathematics, computer science, physics, electrical engineering, neuroscience, and cogni- 
tive science, unified by a common desire to understand the mechanisms for information 
processing in the brain, and to apply this knowledge to the design of new and efficient 
computational and statistical strategies. Annual gatherings with a single track format for 
oral sessions, complemented by very lively posters sessions, have led over the years to the 
development of an interdisciplinary and highly imeractive community, characterized by a 
fruitful exchange of ideas between participants imerested in theory, algorithms, applica- 
tions, and implememations. The conference cominues to grow and evolve in ways that 
reflect the changing needs and imerests of its constituency. A scan of the preceding NIPS 
volumes provides an inspiring survey of the fast-paced developmere of the field over the 
past 12 years; NIPS both reflects and contributes to this vibrant scientific enterprise. 
In addition to the strong program of comributed papers included here, the conference fea- 
tured a lively schedule of invited speakers. Bernardo Huberman of Xerox PARC gave the 
banquet talk on "The Laws of the Web"; additional invited talks were given by Eero Si- 
moncelli of New York University on "Cortical Normalization Models and the Statistics of 
Visual Images," Larry Abbott of Brandeis University on "Temporally Asymmetric Hebbian 
Learning, Spike Timing, and Neural Response Variability," Haim Sompolinsky of Hebrew 
University on "Computation by Cortical Modules," Eugene Chamiak of Brown University 
on "Statistical Natural Language Processing: Better Living Through Floating-point Num- 
bers," and Nell Gershenfeld of MIT on "Things That Think." 
As is traditional, the main conference was followed by two days of imensive workshops on 
a collection of topics that reflect the diversity of NIPS, and preceded by a day of tutorials 
that provide a pedagogical introduction to topics of current imerest. This year we heard 
from William Bialek of NEC Research on "Do We Understand the Neural Code?," Jean- 
Franqois Cardoso of CNRS and ENST, Paris, on "Independem Component Analysis and 
xvi Preface 
Blind Separation of Signals," Henry Markram of the Weizmann Institute on "Neocortical 
Synapses," Peter Bartlett of the Australian National University on "Learning Theory and 
Generalization for Neural Networks and Other Supervised Learning Techniques," Daniel 
Kersten of the University of Minnesota on "Computational Vision: Principles of Percep- 
tual Inference," and Joachim Buhmann of the University of Bonn on "Exploratory Data 
Analysis and Data Visualization." 
The continued success of NIPS is the result of the parallel and distributed effort of many 
devoted individuals. We extend our thanks to the members of the organizing committee, 
the program committee, the publicity committee and the NIPS Foundation board. Special 
gratitude is due to the 191 referees whose integrated efforts led to the superb conference 
program presented here, and to the authors who provided the high quality material upon 
which the program was built. Thanks to the workshop chairs Richard Zemel and Sue 
Becker, tutorials chair Klaus-Robert Miiller, publicity chair Jon Baxter, treasurer Bartlett 
Mel, and local arrangements chair Arun Jagota. A special thanks to Thomas Petsche, who 
in his role of Publication Chair for NIPS*96 developed an excellent and comprehensive 
set of formatting tools for the production of both the Conference Program and the NIPS 
Proceedings. Finally, we are particularly grateful to Rosemary Miller and Kathie Hibbard 
for their stellar work with all aspects of the conference logistics, from registration and 
administration to dinner menus and poster arrangements; we also thank the many student 
volunteers who assisted them on site. Special thanks to Kathleen Burgess for her highly 
efficient and reliable help with the monumental task of processing the submitted papers, 
and to Pamela Hawkins for her invaluable assistance with the coordination of the tasks of 
the program committee. 
We hope that the papers will provide useful and enjoyable reading, and we look forward to 
seeing you all at the many NIPS to come! 
Michael S. Kearns, AT&T Labs- Research 
Sara A. Solla, Northwestern University 
David A. Cohn, Just Research 
January 1999 
NIPS Committees 
Organizing Committee 
General Chair 
Program Chair 
Workshop Co-Chairs 
Tutorials Chair 
Publicity Chair 
Publications Chair 
Treasurer 
Local Arrangements Chair 
Contracts 
Web Master 
Michael S. Kearns, AT&T Labs - Research 
Sara A. Solla, Northwestern University 
Richard Zemel, University of Arizona 
Sue Becker, McMaster University 
Klaus Mueller, GMD FIRST 
Jonathan Baxter, Australian National University 
David A. Cohn, Harlequin, Inc. and Just Research 
Bartlett Mel, University of Southern California 
Arun Jagota, University of California, Santa Cruz 
Scott Kirkpatrick, IBM Watson Labs 
Gerald Tesauro, IBM Watson Labs 
Doug Baker, Carnegie Mellon University 
Program Committee 
Program Chair 
Program Co-chairs 
Sara A. Solla, Northwestern University 
Andrew Barto, University of Massachusetts, Amherst 
Joachim Buhmann, University of Bonn 
Lars Kai Hansen, Danish Technical University 
Nathan Intrator, Tel-Aviv University 
Robert Jacobs, University of Rochester 
Esther Levin, AT&T Labs - Research 
Alexandre Pouget, Georgetown University 
David Saad, Aston University 
Lawrence Saul, AT&T Labs - Research 
Sebastian Thrun, Carnegie Mellon University 
Manfred K. Warmuth, University of California, Santa Cruz 
Yair Weiss, Massachusetts Institute of Technology 
xviii NIPS Committees 
Publicity Committee 
Publicity Chair 
Liaisons 
Australia, Singapore and India 
Europe 
Hong Kong, China and Taiwan 
Israel 
Sapan 
Turkey 
United Kingdom 
South America 
Jonathan Baxter, Australian National University 
Marwan Jabri, University of Sydney 
Joachim Buhmann, University of Bonn 
Lei Xu, Chinese University of Hong Kong 
Hava Siegelmann, Technion 
Kenji Doya, ATR Research Laboratories 
Ethem Alpaydin, Bogazici University 
Alan Murray, Edinburgh University 
Andreas Meier, Simon Bolivar University 
NIPS Foundation Board Members 
President 
Vice President for Development 
Treasurer 
Secretary 
Members 
Emeritus 
NIPS*98 General Chair 
Terrence Sejnowski, The Salk Institute 
Gary Blasdel, Harvard Medical School 
Eric Mjolsness, Jet Propulsion Laboratory 
Gerald Tesauro, IBM Watson Labs 
Leo Breiman, University of California, Berkeley 
Jack Cowan, University of Chicago 
Stephen J. Hanson, Rutgers University 
Michael I. Jordan, University of California, Berkeley 
Scott Kirkpatrick, IBM Watson Labs 
Richard Lippmann, MIT Lincoln Laboratory 
John Moody, Oregon Graduate Institute 
Michael Mozer, University of Colorado, Boulder 
Dave Touretzky, Carnegie Mellon University 
Terrence Fine, Cornell University 
Eve Marder, Brandeis University 
Michael S. Kearns, AT&T Labs - Research 
Reviewers 
Tulay Adali 
Shunichi Amari 
Chuck Anderson 
James Anderson 
Martin Anthony 
Chris Atkeson 
Andrew Back 
Wyeth Bair 
Leemon Baird 
Pierre Baldi 
Dana Ballard 
Shumeet Baluja 
David Barber 
Peter Bartlett 
Eric Baum 
Jonathan Baxter 
Tony Bell 
Yoshua Bengio 
Michael Berry 
Michael Biehl 
Christopher Bishop 
Avrim Blum 
Leon Bottou 
Herve Bourlard 
Craig Boutilier 
Justin Boyan 
Matthew Brand 
Christoph Bregler 
Emanuela Bricolo 
Timothy Brown 
Nicolas Brunel 
Nader Bshouty 
Christopher Burges 
Neil Burgess 
Matteo Carandini 
Richard Camana 
Nestor Caticha 
Gert Cauwenberghs 
Nicolo Cesa-Bianchi 
Eric Chang 
Andrzej Cichocki 
Ton Coolen 
Corinna Cortes 
Gary Cottrell 
Mark Craven 
Trevor Darrell 
Gustavo Deco 
Michael Denham 
Joachim Diederich 
Thomas Dietterich 
Leif Finkel 
Jozsef Fiser 
Paolo Frasconi 
William Freeman 
Brendan Frey 
Nir Friedman 
Juergen Fritsch 
Bernd Fritzke 
Patrick Gallinari 
Zoubin Ghahramani 
C. Lee Giles 
Moises Goldszmidt 
Geoffrey Goodhill 
Michael Gray 
Edwin Hancock 
John Harris 
Simon Haykin 
David Heckerman 
John Hertz 
Tom Heskes 
Thomas Hofmann 
David Horn 
Yu Hen Hu 
Michael Isard 
Tommi Jaakkola 
Marwan Jabri 
Charles Jankowski 
Nathalie Japkowicz 
Thorsten Joachims 
Michael Jordan 
Yoshiyuki Kabashima 
Leslie Kaelbling 
Bert Kappen 
Shigeru Katagiri 
Mitsuo Kawato 
Daniel Kersten 
Hiroaki Kitano 
Adam Kowalczyk 
Stefan Kremer 
John Kruschke 
Tor Sverre Lande 
Jan Larsen 
Peter Latham 
John Lazzaro 
Yann LeCun 
Daniel Lee 
Tai-Sing Lee 
Te-Won Lee 
Todd Leen 
Torsten Lehmann 
Nick Littlestone 
Michael Littman 
David Lowe 
Gabor Lugosi 
Sridhar Mahadevan 
Maja Mataric 
Andrew McCallum 
Ron Meir 
Bartlett Mel 
David Miller 
Kenneth Miller 
Melanie Mitchell 
Tom Mitchell 
Baback Moghaddam 
Andrew Moore 
Javier Movellan 
Michael Mozer 
Klaus Mueller 
Paul Munro 
Noboru Murata 
Alan Murray 
Rod Murray-Smith 
Jean-Pierre Nadal 
Kenny Ng 
Klaus Obermayer 
Erkki Oja 
John Oliensis 
Bruno Olshausen 
Manfred Opper 
xx Reviewers 
Guenther Palm 
Ronald Parr 
Barak Pearlmutter 
Fernando Pereira 
Michael Perrone 
Thomas Petsche 
Roberto Pieraccini 
Jose Principe 
Mazin Rahim 
Anand Rangarajan 
Rajesh Rao 
Carl Rasmussen 
A. David Redish 
Pamela Reinagel 
Thomas Richardson 
Tony Robinson 
Dan Roth 
Sam Roweis 
Henry Rowley 
Philip Sabes 
Maneesh Sahani 
Eric Saund 
Stefan Schaal 
Rob Schapire 
Jeff Schneider 
Bernhard Sch61kopf 
Holger Schwenk 
Sebastian Seung 
Jude Shavlik 
John Shawe-Taylor 
Patrice Simard 
Eero Simoncelli 
Yoram Singer 
Satinder Singh 
Padhraic Smyth 
Peter Sollich 
Michael Spivey 
Richard Sutton 
Csaba Szepesvari 
Michiaki Taniguchi 
Josh Tenenbaum 
Michael Tipping 
Naftali Tishby 
Mike Titterington 
Geoffrey Towell 
Volker Tresp 
Todd Troyer 
Ian Underwood 
Chris van den Broeck 
Marc Van Hulle 
Nuno Vasconcelos 
Paul Viola 
Vladimir Vovk 
Yuedong Wang 
Daphna Weinshall 
Wim Wiegerinck 
Christopher Williams 
Ronald Williams 
Laurenz Wiskott 
Robin Woodbum 
Richard Zemel 
Kechen Zhang 
PART I 
COGNITIVE SCIENCE 
