Paper Abstract and Keywords |
Presentation |
2012-06-19 10:00
A study on an optimization algorithm for semi-supervised SVM using parametric programing Kohei Ogawa, Ichiro Takeuchi (NIT), Masashi Sugiyama (Tokyo Tech) IBISML2012-1 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
The goal of semi-supervised learning is to incorporate unlabeled instances as well as labeled ones for improving classification performance to unseen data.
Semi-supervised SVM
(S3VM) is an SVM-like classification algorithm for semi-supervised learning scenario.
The problem of learning S3VM is formulated either as a hard combinatorial optimization problem or a non-convex optimization problem.
Since the global optimal solution of ${\rm S^3VM}$ is difficult to find except for small problems,
the practical goal of S3VM studies is to develop an algorithm that can find good local optimal solutions using heuristic search techniques such as deterministic annealing.
In this report we introduce a parametric programming approach to
S3VM.
The main contribution of this paper is to derive the necessary and sufficient conditions of a local optimal solution of S3VM,
and to develop an algorithm that can compute a path of local optimal solutions that always satisfy the local optimality conditions
when the effect of unlabeled instances are gradually increased.
The proposed algorithm empirically produces better local optimal solutions possibly because homotopy approach offers similar advantage as deterministic annealing with infinitely small steps.
Moreover, our approach allows stable
and efficient model selection for controlling the influence of unlabeled instances .
We experimentally demonstrate the usefulness of the proposed method. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Semi-Supervised learning / Semi-Supervised SVM / Parametric Programing / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 112, no. 83, IBISML2012-1, pp. 1-8, June 2012. |
Paper # |
IBISML2012-1 |
Date of Issue |
2012-06-12 (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) |
Notes on Review |
This article is a technical report without peer review, and its polished version will be published elsewhere. |
Download PDF |
IBISML2012-1 |
Conference Information |
Committee |
IBISML |
Conference Date |
2012-06-19 - 2012-06-20 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
Campus plaza Kyoto |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
General topics on machine learning and its application |
Paper Information |
Registration To |
IBISML |
Conference Code |
2012-06-IBISML |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
A study on an optimization algorithm for semi-supervised SVM using parametric programing |
Sub Title (in English) |
|
Keyword(1) |
Semi-Supervised learning |
Keyword(2) |
Semi-Supervised SVM |
Keyword(3) |
Parametric Programing |
Keyword(4) |
|
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Kohei Ogawa |
1st Author's Affiliation |
Nagoya Institute of Technology (NIT) |
2nd Author's Name |
Ichiro Takeuchi |
2nd Author's Affiliation |
Nagoya Institute of Technology (NIT) |
3rd Author's Name |
Masashi Sugiyama |
3rd Author's Affiliation |
Tokyo Institute of Technology (Tokyo Tech) |
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 |
() |
19th Author's Name |
|
19th Author's Affiliation |
() |
20th Author's Name |
|
20th Author's Affiliation |
() |
Speaker |
Author-1 |
Date Time |
2012-06-19 10:00:00 |
Presentation Time |
30 minutes |
Registration for |
IBISML |
Paper # |
IBISML2012-1 |
Volume (vol) |
vol.112 |
Number (no) |
no.83 |
Page |
pp.1-8 |
#Pages |
8 |
Date of Issue |
2012-06-12 (IBISML) |