Presentation | 2007-12-22 Rank Reduction by Cross-Validated Backpropagation Masashi SEKINO, Katsumi NITTA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In this paper, we propose cross-validated backpropagation for training neural networks. The experimental results show that the average generalization error of the proposed method is smaller than those of the ordinal backpropagation, early-stopping and Bayes estimation, that the proposed method gives the same results as those by large rank models and small rank models when the rank of the true function is small, that the plateau of the learning, which is observed, when the backpropagation is applied, , is also observed when the proposed method is applied, and that the proposed method also gives good approximation performance when the true function is almost unidentifiable. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Neural Network / Backpropagation / Overfitting / Cross-Validation / Reduced Rank Regression |
Paper # | NC2007-76 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 2007/12/15(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) | Rank Reduction by Cross-Validated Backpropagation |
Sub Title (in English) | |
Keyword(1) | Neural Network |
Keyword(2) | Backpropagation |
Keyword(3) | Overfitting |
Keyword(4) | Cross-Validation |
Keyword(5) | Reduced Rank Regression |
1st Author's Name | Masashi SEKINO |
1st Author's Affiliation | () |
2nd Author's Name | Katsumi NITTA |
2nd Author's Affiliation | Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology |
Date | 2007-12-22 |
Paper # | NC2007-76 |
Volume (vol) | vol.107 |
Number (no) | 410 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |