| 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 | 2022-09-15 15:05 Improving Efficiency of Regularization Path Computation in Safe Pattern Pruning via Multiple Referential Solutions Takumi Yoshida (Nitech), Hiroyuki Hanada (RIKEN), Kazuya Nakagawa, Shinya Suzumura, Onur Boyar, Kazuki Iwata (Nitech), Shun Shimura, Yuji Tanaka (NaogyaU), Masayuki Karasuyama (Nitech), Kouichi Taji (NaogyaU), Koji Tsuda (UTokyo/RIKEN), Ichiro Takeuchi (NaogyaU/RIKEN)  IBISML2022-38 |
|
| Abstract | (in Japanese) | (See Japanese page) |
| (in English) | Safe Screening and Safe Pattern Pruning are methods for efficiently modeling high-dimensional features by $L_1$-regularized learning. These methods improve the efficiency of it by identifying features to be excluded from the model in advance. These methods require a feasible solution to be used as a reference, and it is known that in $L_1$ regularization, the training result with the previous regularization parameter can be used as a reference solution. In this study, we focus on the situations where multiple reference solutions are available, such as Elastic Net regularization. We propose to use these multiple reference solutions in such cases. Then we demonstrate that stronger screening is possible in real data experiments. | |
| Keyword | (in Japanese) | (See Japanese page) |
| (in English) | L1-Regularized Optimization / Safe Screening / Safe Pattern Pruning / Pattern Mining / Structured Input / Convex Optimization / / | |
| Reference Info. | IEICE Tech. Rep., vol. 122, no. 189, IBISML2022-38, pp. 39-46, Sept. 2022. | |
| Paper # | IBISML2022-38 | |
| Date of Issue | 2022-09-08 (IBISML) | |
| ISSN | 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 | IBISML2022-38 |
|
| Conference Information | |
| Committee | IBISML |
| Conference Date | 2022-09-15 - 2022-09-15 |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Keio Univ. (Yagami Campus) |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Machine Learning, etc. |
| Paper Information | |
| Registration To | IBISML |
| Conference Code | 2022-09-IBISML |
| Language | Japanese |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Improving Efficiency of Regularization Path Computation in Safe Pattern Pruning via Multiple Referential Solutions |
| Sub Title (in English) | |
| Keyword(1) | L1-Regularized Optimization |
| Keyword(2) | Safe Screening |
| Keyword(3) | Safe Pattern Pruning |
| Keyword(4) | Pattern Mining |
| Keyword(5) | Structured Input |
| Keyword(6) | Convex Optimization |
| Keyword(7) | |
| Keyword(8) | |
| 1st Author's Name | Takumi Yoshida |
| 1st Author's Affiliation | Nagoya Institute of Technology (Nitech) |
| 2nd Author's Name | Hiroyuki Hanada |
| 2nd Author's Affiliation | Institute of Physical and Chemical Research (RIKEN) |
| 3rd Author's Name | Kazuya Nakagawa |
| 3rd Author's Affiliation | Nagoya Institute of Technology (Nitech) |
| 4th Author's Name | Shinya Suzumura |
| 4th Author's Affiliation | Nagoya Institute of Technology (Nitech) |
| 5th Author's Name | Onur Boyar |
| 5th Author's Affiliation | Nagoya Institute of Technology (Nitech) |
| 6th Author's Name | Kazuki Iwata |
| 6th Author's Affiliation | Nagoya Institute of Technology (Nitech) |
| 7th Author's Name | Shun Shimura |
| 7th Author's Affiliation | Nagoya University (NaogyaU) |
| 8th Author's Name | Yuji Tanaka |
| 8th Author's Affiliation | Nagoya University (NaogyaU) |
| 9th Author's Name | Masayuki Karasuyama |
| 9th Author's Affiliation | Nagoya Institute of Technology (Nitech) |
| 10th Author's Name | Kouichi Taji |
| 10th Author's Affiliation | Nagoya University (NaogyaU) |
| 11th Author's Name | Koji Tsuda |
| 11th Author's Affiliation | The University Tokyo/Institute of Physical and Chemical Research (UTokyo/RIKEN) |
| 12th Author's Name | Ichiro Takeuchi |
| 12th Author's Affiliation | Nagoya University/Institute of Physical and Chemical Research (NaogyaU/RIKEN) |
| Speaker | Author-1 |
| Date Time | 2022-09-15 15:05:00 |
| Presentation Time | 25 minutes |
| Registration for | IBISML |
| Paper # | IBISML2022-38 |
| Volume (vol) | vol.122 |
| Number (no) | no.189 |
| Page | pp.39-46 |
| #Pages | 8 |
| Date of Issue | 2022-09-08 (IBISML) |
[Return to Top Page]