Abstract | ||
---|---|---|
We discuss effects of realistically dilute connectivity on a neural network previously proposed as a model of hippocampal associative memory. Several criteria for setting neuronal activation thresholds are studied. Even with optimal thresholding, dilution induces a substantial information loss in stored patterns compared to presented patterns. Consequently, a stricter constraint than previous ones arises on the model's storage capacity. Furthermore, we argue that such constraints depend sensitively on the specific, subjective criteria chosen for storage quality. We thus propose that additional performance measures be considered. In particular, the relationship between firing rates of original and attractor patterns is discussed. |
Year | DOI | Venue |
---|---|---|
1999 | 10.1016/S0925-2312(99)00048-X | Neurocomputing |
Keywords | Field | DocType |
Attractor neural networks,Associative memory,Hippocampus | Attractor,Information loss,Content-addressable memory,Pattern recognition,Bidirectional associative memory,Artificial intelligence,Thresholding,Artificial neural network,Hippocampal formation,Hippocampus,Mathematics,Machine learning | Journal |
Volume | ISSN | Citations |
26-27 | 0925-2312 | 0 |
PageRank | References | Authors |
0.34 | 2 | 1 |
Name | Order | Citations | PageRank |
---|---|---|---|
Miguel Maravall | 1 | 5 | 1.58 |