Learning in a two-layer neural network of edge detectors

H. Sompolinsky*, N. Tishby

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

17 Scopus citations

Abstract

Learning from examples to count domains in one-dimensional patterns is studied. Increasing the number of examples used for training a network to perform the task is equivalent to the annealing of a one-dimensional Ising model. The generalization error falls off exponentially with the number of examples per weight. The related contiguity problem, where the network discriminates between patterns with small and large number of domains, exhibits a first-order phase transition to perfect generalization at all temperatures. Monte Carlo simulations of both models are in very good agreement with the theoretical predictions.

Original languageEnglish
Pages (from-to)567-572
Number of pages6
JournalLettere Al Nuovo Cimento
Volume13
Issue number6
DOIs
StatePublished - 15 Nov 1990

Fingerprint

Dive into the research topics of 'Learning in a two-layer neural network of edge detectors'. Together they form a unique fingerprint.

Cite this