| Presentation | 1998/3/19 Information geometrical theory of mean field approximation Toshiyuki Tanaka, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Mean field approximation, a method originally developed in statistical physics, provides a framework of deterministically approximating stochasticity of neural networks, and has been successfully applied to problems of optimization and learning. However, there are few information-theoretic studies as to the properties of the approximation, except for the so-called naive mean field approximation. In this study I show a general information-theoretic formulation of mean field approximation based on information geometry. I also show that from this formulation one can derive not only the naive mean field theory but also TAP approach, which is well known in statistical physics as giving more accurate approximations, and the linear response theorem, which allows treatment of statistical correlations within the framework of mean field approximation, in a natural and consistent way. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | mean field approximation / information geometry / perturbation expansion / learning / neural network |
| Paper # | |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 1998/3/19(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) | Information geometrical theory of mean field approximation |
| Sub Title (in English) | |
| Keyword(1) | mean field approximation |
| Keyword(2) | information geometry |
| Keyword(3) | perturbation expansion |
| Keyword(4) | learning |
| Keyword(5) | neural network |
| 1st Author's Name | Toshiyuki Tanaka |
| 1st Author's Affiliation | Graduate School of Electrical Engineering, Tokyo Metropolitan University() |
| Date | 1998/3/19 |
| Paper # | |
| Volume (vol) | vol.97 |
| Number (no) | 623 |
| Page | pp.pp.- |
| #Pages | 8 |
| Date of Issue | |