Abstract
We study three ways of processing of temporal sequences of patterns stored in a neural network; retrieval, counting, and recognition. These three processes are studied analytically in a strongly diluted neural network. The results compare qualitatively to simulations in fully connected networks.
Original language | English |
---|---|
Pages (from-to) | 235-238 |
Number of pages | 4 |
Journal | Physical Review Letters |
Volume | 61 |
Issue number | 2 |
DOIs | |
State | Published - 1988 |