TCB Publications - Abstract

Joachim Buhmann, Robert Divko, and Klaus Schulten. On sparsely coded associative memories. In L. Personnaz and G. Dreyfus, editors, Neural Networks: From Models to Applications, N'EURO '88, pp. 360-371. EZIDET, Paris, 1989.

BUHM89A A class of neural networks with adaptive threshold and global inhibitory interactions is proposed. The networks are capable of nearly optimal storage of sparsely coded patterns, i.e. patterns with low level of activity. We present a replica symmetric solution of the mean field equations for the noise free case. The network shows the following, remarkable properties: 1) For low level of activity a the storage capacity increases as -[a 1n a]-1; up to 0.38 bits/synapse can be stored. The network capacity approaches the theoretical upper bounds derived by Gardner (1988). 2) Spurious states representing superpositions of stored patterns can be suppressed by global inhibition; 3) The network is not opinionated, i.e. by assuming a state of low activity it categorizes respective inputs as not similar enough to any patterns stored.

Download Full Text

The manuscripts available on our site are provided for your personal use only and may not be retransmitted or redistributed without written permissions from the paper's publisher and author. You may not upload any of this site's material to any public server, on-line service, network, or bulletin board without prior written permission from the publisher and author. You may not make copies for any commercial purpose. Reproduction or storage of materials retrieved from this web site is subject to the U.S. Copyright Act of 1976, Title 17 U.S.C.

Download full text: PDF (427.5KB)