Isaac Newton Institute for Mathematical Sciences

Neural Networks and Machine Learning

July - December 1997

Organisers: C M Bishop (Aston), D Haussler (UCSC), G E Hinton (Toronto), M Niranjan (Cambridge), L G Valiant (Harvard)

A Newton Institute Themed Week

ON-LINE LEARNING IN NEURAL NETWORKS

17 - 21 November, 1997

Isaac Newton Institute, Cambridge, U.K.

Organisers: D Saad (Aston) and C M Bishop (Microsoft)

Programme (Abstracts are available for each talk)

Monday 17 November

11:30 - 12:15 Chris Van den Broeck (Limburgs, Belgium)
Off-line Versus On-line Learning in an Unsupervised Scenario

12:30 - 14:30 Lunch

14:30 - 15:15 Leon Bottou (AT&T, USA)
Stochastic Approximations and Online Learning

15:30 - 16:00 Tea

16:00 - 16:45 Todd Leen (OGI, USA)
1. A Perturbation Expansion for Weight Space Probability Densities (with Genevieve Orr)

2.
Ensemble Dynamics of Stochastic Approximation: An Exactly Solvable Model (with John E. Moody)

17:30 Wine Reception at the Newton Institute

Tuesday 18 November

10:00 - 10:45 Noboru Murata (RIKEN, Japan)
Statistical Analysis of On-line Learning

11:00 - 11:30 Coffee

11:30 - 12:15 Klaus-Robert Müller (GMD FIRST, Berlin), Noboru Murata, Andreas Ziehe and Shun-ichi Amari
Applying On-line Learning to Non-stationary Blind Separation

12:30 - 14:00 Lunch

14:30 - 15:15 Siegfried Boes and Shun-ich Amari (RIKEN, Japan)
The Role of the Learning Rate in On-line Learning

15.30 - 16:00 Tea

16:00 - 16:45 Magnus Rattray and David Saad (Aston University, UK)
Optimal On-line Learning for Multilayer Neural Networks

Wednesday 19 November

10:00 - 10:45 David Saad and Magnus Rattray (Aston University, UK)
Incorporating Curvature Information in On-Line Learning

11:00 - 11:30 Coffee

11:30 - 12:15 Manfred Opper (Aston University, UK)
A Bayesian Approach to Online Learning

12:30 - 14:30 Lunch

14:30 - 15:15 Sara A. Solla (Northwestern) and Ole Winther (CONNECT)
Optimal Bayesian On-line Learning

15:00 - 15:30 Tea

15:30 - 16:15 Luis B. Almeida (Lisbon, Portugal)
Adapting Parameters in On-line Learning

Thursday 20 November

10:00 - 10:45 Enno Schlösser and Michael Biehl (Wurzburg, Germany)
The Dynamics of On-line Principal Component Analysis

11:00 - 11:30 Coffee

11:30 - 12:15 Mauro Copelli (Limburgs, Belgium), M. Biehl, N. Caticha, O. Kinouchi, R. Eichorn, R. Simonetti and P. Riegler,
Noise Robustness in Boolean Machines

12:30 - 14:30 Lunch

14:30 - 15:15 Yoshiyuki Kabashima (Tokyo, Japan)
Learning a Decision Boundary from Stochastic Examples: Incremental Algorithms with and without Queries

15:30 - 16:00 Tea

16:00 - 16:45 Tom Heskes (Nijmegen, The Netherlands)
On-line Learning with Correlated Patterns

Friday 21 November

10:00 - 10:45 David Barber (Nijmegen) and Peter Sollich (Edinburgh)
Online Learning from Finite Training Sets: Soft Committee Machine

11:00 - 11:30 Coffee

11:30 - 12:15 ACC Coolen (KCL, UK) and David Saad (Aston, UK)
Theory of On-Line Learning with Restricted Training Sets

12:30 Lunch and close


This workshop will form a component of the Newton Institute programme on Neural Networks and Machine Learning, organised by C M Bishop, D Haussler, G E Hinton, M Niranjan and L G Valiant.

Copyright © Isaac Newton Institute