| IEICE Technical Committee Submission System Conference Paper's Information |
Online Proceedings [Sign in] Tech. Rep. Archives |
| Go Top Page | Go Previous | [Japanese] / [English] |
| Paper Abstract and Keywords | ||
| Presentation | 2011-10-20 13:10 Statistical Mechanics of Node-Perturbation Learning for Nonlinear Perceptron Kazuyuki Hara (Nihon Univ.), Kentaro Katahira (JST), Kazuo Okanoya (RIKEN), Masato Okada (Tokyo Univ.)  NC2011-63 |
|
| Abstract | (in Japanese) | (See Japanese page) |
| (in English) | Node-perturbation learning is a kind of statistical gradient descent algorithm that can be applied to problems where the objective function is not explicitly formulated, including reinforcement learning. In the previous report, we explored statistical mechanics of node-perturbation learning with M linear perceptrons, and analyzed the generalization. From the results, we clarified that the cross-take noise enlarges the generalization error of the output when there are more outputs. In this report, we propose a novel node-perturbation learning for non-linear perceptron, and analyze the behavior by using statistical mechanics methods. From the results, we found that the signal part and cross-talk part of the output error are different forms when non-linear perceptrons are used, and cross-talk part is non-monotonically increased for number of outputs. Moreover, increase of the generalization error for more outputs is held down using non-linear perceptron compare to that using linear perceptrons. |
|
| Keyword | (in Japanese) | (See Japanese page) |
| (in English) | Node-perturbation learning / Non-linear perceptron / on-line learning / generalization error / statistical mechanics method / / / | |
| Reference Info. | IEICE Tech. Rep., vol. 111, no. 241, NC2011-63, pp. 107-112, Oct. 2011. | |
| Paper # | NC2011-63 | |
| Date of Issue | 2011-10-12 (NC) | |
| ISSN | Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 | |
| Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| Download PDF | NC2011-63 |
|
| Conference Information | |
| Committee | NC |
| Conference Date | 2011-10-19 - 2011-10-20 |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Ohashi Campus, Kyushu Univ. |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Implementation of neurocomputing, modeling for human science, and general |
| Paper Information | |
| Registration To | NC |
| Conference Code | 2011-10-NC |
| Language | Japanese |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Statistical Mechanics of Node-Perturbation Learning for Nonlinear Perceptron |
| Sub Title (in English) | |
| Keyword(1) | Node-perturbation learning |
| Keyword(2) | Non-linear perceptron |
| Keyword(3) | on-line learning |
| Keyword(4) | generalization error |
| Keyword(5) | statistical mechanics method |
| Keyword(6) | |
| Keyword(7) | |
| Keyword(8) | |
| 1st Author's Name | Kazuyuki Hara |
| 1st Author's Affiliation | Nihon University (Nihon Univ.) |
| 2nd Author's Name | Kentaro Katahira |
| 2nd Author's Affiliation | Japan Science Technology Agency, ERATO Okanoya Emotional Information Project (JST) |
| 3rd Author's Name | Kazuo Okanoya |
| 3rd Author's Affiliation | RIKEN (RIKEN) |
| 4th Author's Name | Masato Okada |
| 4th Author's Affiliation | The University of Tokyo (Tokyo Univ.) |
| 5th Author's Name | |
| 5th Author's Affiliation | () |
| 6th Author's Name | |
| 6th Author's Affiliation | () |
| 7th Author's Name | |
| 7th Author's Affiliation | () |
| 8th Author's Name | |
| 8th Author's Affiliation | () |
| 9th Author's Name | |
| 9th Author's Affiliation | () |
| 10th Author's Name | |
| 10th Author's Affiliation | () |
| 11th Author's Name | |
| 11th Author's Affiliation | () |
| 12th Author's Name | |
| 12th Author's Affiliation | () |
| 13th Author's Name | |
| 13th Author's Affiliation | () |
| 14th Author's Name | |
| 14th Author's Affiliation | () |
| 15th Author's Name | |
| 15th Author's Affiliation | () |
| 16th Author's Name | |
| 16th Author's Affiliation | () |
| 17th Author's Name | |
| 17th Author's Affiliation | () |
| 18th Author's Name | |
| 18th Author's Affiliation | () |
| Speaker | Author-1 |
| Date Time | 2011-10-20 13:10:00 |
| Presentation Time | 25 minutes |
| Registration for | NC |
| Paper # | NC2011-63 |
| Volume (vol) | vol.111 |
| Number (no) | no.241 |
| Page | pp.107-112 |
| #Pages | 6 |
| Date of Issue | 2011-10-12 (NC) |
[Return to Top Page]