Training a perceptron in a discrete weight space

M. Rosen-Zvi, I. Kanter

Research output: Contribution to journalArticlepeer-review

Abstract

Learning in a perceptron in a discrete weight space is examined analytically and numerically. The learning algorithm is based on the training of the continuous perceptron and prediction following the clipped weights. The realtions between the overlaps of the continuous teacher with the discrete/continuous students were derived. The results show that learning in the case of finite depth is possible by using a continuous precursors.
Original languageEnglish
JournalPhysical Review E - Statistical, Nonlinear, and Soft Matter Physics
Volume64
Issue number4 II
StatePublished - 1 Oct 2001

Fingerprint

Dive into the research topics of 'Training a perceptron in a discrete weight space'. Together they form a unique fingerprint.

Cite this