On-line learning from restricted training sets in multilayer neural networks

A.C.C. Coolen, D. Saad, Yuan-Sheng Xiong

Research output: Contribution to journalArticlepeer-review


We analyse the dynamics of on-line learning in multilayer neural networks where training examples are sampled with repetition and where the number of examples scales with the number of network weights. The analysis is based on monitoring a set of macroscopic variables from which the training and generalisation errors can be calculated. A closed set of dynamical equations is derived using the dynamical replica method and is solved numerically. The theoretical results are consistent with those obtained by computer simulations.
Original languageEnglish
Pages (from-to)691-697
Number of pages7
JournalEurophysics Letters
Issue number6
Publication statusPublished - 1 Sept 2000

Bibliographical note

Funding: DS and YX acknowledge support by EPSRC (GR/L52093) and the British Council (ARC1037).


Dive into the research topics of 'On-line learning from restricted training sets in multilayer neural networks'. Together they form a unique fingerprint.

Cite this