Presentation | 2013-07-19 Statistical Mechanics of node-perturbation Learning using two independent noises Kazuyuki HARA, Kentaro KATAHIRA, Masato OKADA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Node perturbation learning is a stochastic gradient descent method for neural networks. It estimates the gradient by comparing an evaluation of the perturbed output and the unperturbed output performance, which we call the baseline. However, the unperturbed performance is a difficult setting. Cho et al. proposed node-perturbation using two independent noises, and it is tractable for real problem. In this report, we analyze the node-perturbation learning using two independent noises by using statistical mechanics methods. From the results, we find that the residual error becomes small compared to the original one when the variance of two noises is the same. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | node-perturbation learning / two independent noises / on-line learning / generalization error / statistical mechanics method |
Paper # | NC2013-17 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 2013/7/12(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) | Statistical Mechanics of node-perturbation Learning using two independent noises |
Sub Title (in English) | |
Keyword(1) | node-perturbation learning |
Keyword(2) | two independent noises |
Keyword(3) | on-line learning |
Keyword(4) | generalization error |
Keyword(5) | statistical mechanics method |
1st Author's Name | Kazuyuki HARA |
1st Author's Affiliation | College of Industrial Engineering, Nihon University() |
2nd Author's Name | Kentaro KATAHIRA |
2nd Author's Affiliation | Department of Life Sciences, Graduate School of Arts and Sciences, The University of Tokyo:Riken |
3rd Author's Name | Masato OKADA |
3rd Author's Affiliation | Graduate School of Frontier Sciences, The University of Tokyo:Riken |
Date | 2013-07-19 |
Paper # | NC2013-17 |
Volume (vol) | vol.113 |
Number (no) | 148 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |