A trade-off formula in designing asymmetric neural networks (Q2857971)
From MaRDI portal
 | This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use this page instead for the normal view: A trade-off formula in designing asymmetric neural networks |
scientific article; zbMATH DE number 6229159
| Language | Label | Description | Also known as |
|---|---|---|---|
| English | A trade-off formula in designing asymmetric neural networks |
scientific article; zbMATH DE number 6229159 |
Statements
A trade-off formula in designing asymmetric neural networks (English)
0 references
19 November 2013
0 references
asymetric neural networks
0 references
storage capacity
0 references
synaptic couplings
0 references
0 references
0.85575885
0 references
0.8523926
0 references
0.8502566
0 references
0 references
A two-state bipolar asymmetric neural network of \(N\) neurons is considered. The given memory set will be stored by means of the so-called deterministic successive over-relaxation (SOR) rule. After a short investigation on the eigenvalue distribution of the synaptic neural network couplings, a formula connecting the symmetric degree of the synaptic couplings with the storage capacity and the designing parameter \(k\) is proposed. The parameter \(k\) states the number of patterns that could be stored as memories. Some examples are used to justify this formula.
0 references
