Presentation | 1996/5/24 Probabilistic Memory Capacity of Recurrent Neural Networks Seiji MIYOSHI, Kenji NAKAYAMA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In this paper, we investigate the upper bound of the memory capacity when recurrent neural networks(RNN) are used as associative memories. When an RNN with no self-feedback is used as an associative memory, whether a certain pattern set can be memorized or not depends on the combination of patterns to be memorized. So, in this paper, the idea "probabilistic memory capacity" is introduced. The memory capacity of an RNN is estimated by the probability that the given number of random patterns can be memorized at equilibrium states. The way to apply the linear programming method in order to calculate the probabilistic memory capacity is described. The probabilistic memory capacities about some numbers of units are shown. Finally, the properties of the probabilistic memory capacity when the number of units is large are estimated. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | recurrent neural network / associative memory / memory capacity / probabilistic memory capacity / linear programming method |
Paper # | NC96-5 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 1996/5/24(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Neurocomputing (NC) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Probabilistic Memory Capacity of Recurrent Neural Networks |
Sub Title (in English) | |
Keyword(1) | recurrent neural network |
Keyword(2) | associative memory |
Keyword(3) | memory capacity |
Keyword(4) | probabilistic memory capacity |
Keyword(5) | linear programming method |
1st Author's Name | Seiji MIYOSHI |
1st Author's Affiliation | Graduate School of Natural Science and Technology, Kanazawa Univ.:Department of Electronic Eng., Kobe City College of Technology() |
2nd Author's Name | Kenji NAKAYAMA |
2nd Author's Affiliation | Department of Electrical and Computer Eng. Faculty of Eng., Kanazawa Univ. |
Date | 1996/5/24 |
Paper # | NC96-5 |
Volume (vol) | vol.96 |
Number (no) | 76 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |