Summary
International Symposium on Nonlinear Theory and its Applications
2005
Session Number:1-2-3
Session:
Number:1-2-3-2
Associative Memory Learns Markov Patterns Better than I.I.D. Patternes in Asynchronously Overlapping Networks.
Yutaka Jitsumatsu, Shota Inoue, Tohru Kohda,
pp.234-237
Publication Date:2005/10/18
Online ISSN:2188-5079
Summary:
The analogy between the mathematical models of neural networks and code division multiple access (CDMA) systems have been discussed by many papers. Most of them assumed synchronous CDMA. We propose an associative memory described by asynchronous CDMA communications and analyze its performance. Interestingly, input patterns generated by a Markov chain can be memorized in our associative memory model more than independent and identically distributed (i.i.d.) patterns.