| 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 | 2015-06-25 10:20 e-Bagging: The Information Geometric Dual of Breiman's Bagging -- An Application to the Nadaraya-Watson Regression with the k-Nearest Neighbor Crossover Kernel -- Naoki Hamada, Hiroyuki Higuchi, Katsumi Homma (Fujitsu Labs.)  IBISML2015-23 |
|
| Abstract | (in Japanese) | (See Japanese page) |
| (in English) | The $k$-nearest neighbor crossover kernel, which we proposed recently, is a very flexible kernel that is virtually equivalent to using variable matrix bandwidths. Its large degree of freedom carries, one hand, higher prediction accuracy than conventional kernels and, on the other hand, intractable computation cost for kernel function values. Therefore, its application is limited to the tasks such as re-sampling that can be accomplished without computing kernel function values, excluding the tasks such as regression that requires kernel function values. The computation cost mainly comes from the bagging used in this kernel that expresses the kernel function by the arithmetic mean of enormous multivariate Gaussian densities, i.e., a huge Gaussian mixture. In order to reduce the computation cost while still enjoying the bagging effect, this report proposes $e$-bagging that aggregates densities by $e$-mean. We also discuss the duality between the $e$-bagging and the arithmetic bagging from the information geometric standpoint. We show the effectiveness of the $e$-bagging by applying it to the Nadaraya-Watson regression with the $k$-nearest neighbor crossover kernel. | |
| Keyword | (in Japanese) | (See Japanese page) |
| (in English) | bagging / $e$-mean / information geometry / crossover kernel / Nadaraya-Watson regression / / / | |
| Reference Info. | IEICE Tech. Rep., vol. 115, no. 112, IBISML2015-23, pp. 187-194, June 2015. | |
| Paper # | IBISML2015-23 | |
| Date of Issue | 2015-06-16 (IBISML) | |
| 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 | IBISML2015-23 |
|
| Conference Information | |
| Committee | NC IPSJ-BIO IBISML IPSJ-MPS |
| Conference Date | 2015-06-23 - 2015-06-25 |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Okinawa Institute of Science and Technology |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Machine Learning Approach to Biodata Mining, and General |
| Paper Information | |
| Registration To | IBISML |
| Conference Code | 2015-06-NC-BIO-IBISML-MPS |
| Language | Japanese |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | e-Bagging: The Information Geometric Dual of Breiman's Bagging |
| Sub Title (in English) | An Application to the Nadaraya-Watson Regression with the k-Nearest Neighbor Crossover Kernel |
| Keyword(1) | bagging |
| Keyword(2) | $e$-mean |
| Keyword(3) | information geometry |
| Keyword(4) | crossover kernel |
| Keyword(5) | Nadaraya-Watson regression |
| Keyword(6) | |
| Keyword(7) | |
| Keyword(8) | |
| 1st Author's Name | Naoki Hamada |
| 1st Author's Affiliation | Fujitsu Laboratories Ltd. (Fujitsu Labs.) |
| 2nd Author's Name | Hiroyuki Higuchi |
| 2nd Author's Affiliation | Fujitsu Laboratories Ltd. (Fujitsu Labs.) |
| 3rd Author's Name | Katsumi Homma |
| 3rd Author's Affiliation | Fujitsu Laboratories Ltd. (Fujitsu Labs.) |
| 4th Author's Name | |
| 4th Author's Affiliation | () |
| 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 | 2015-06-25 10:20:00 |
| Presentation Time | 25 minutes |
| Registration for | IBISML |
| Paper # | IBISML2015-23 |
| Volume (vol) | vol.115 |
| Number (no) | no.112 |
| Page | pp.187-194 |
| #Pages | 8 |
| Date of Issue | 2015-06-16 (IBISML) |
[Return to Top Page]