Presentation | 1999/10/21 Regularization Term of Multi Layer Perceptrons Masato UCHIDA, Hiroyuki SHIOYA, Tsutomu DA-TE, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Learning methods adding a regularization term to a loss function to be minimized has been proposed for training multi layer perceptrons, in order to avoid overfitting, improve predictive performance, or perform structurization. In this paper, a learning algoithm with non-Bayesian regularization term whose scaduling parameter doesn't become a constant is derived based on a framework of the parametric estimation. The proposed model is remarkable becase the reguralization term can be regarded as a variance of the model if quadratic loss function is used as a loss function. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Regularization term / learning of multi layer perceptrons |
Paper # | NC99-44 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 1999/10/21(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) | Regularization Term of Multi Layer Perceptrons |
Sub Title (in English) | |
Keyword(1) | Regularization term |
Keyword(2) | learning of multi layer perceptrons |
1st Author's Name | Masato UCHIDA |
1st Author's Affiliation | Faculty of Engineering, Hokkaido University() |
2nd Author's Name | Hiroyuki SHIOYA |
2nd Author's Affiliation | Faculty of Engineering, Hokkaido University |
3rd Author's Name | Tsutomu DA-TE |
3rd Author's Affiliation | Faculty of Engineering, Hokkaido University |
Date | 1999/10/21 |
Paper # | NC99-44 |
Volume (vol) | vol.99 |
Number (no) | 382 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |