Efficient reformulation of 1-norm ranking SVM

研究成果: ジャーナルへの寄稿記事

抄録

Finding linear functions that maximize AUC scores is important in ranking research. A typical approach to the ranking problem is to reduce it to a binary classification problem over a new instance space, consisting of all pairs of positive and negative instances. Specifically, this approach is formulated as hard or soft margin optimization problems over pn pairs of p positive and n negative instances. Solving the optimization problems directly is impractical since we have to deal with a sample of size pn, which is quadratically larger than the original sample size p + n. In this paper, we reformulate the ranking problem as variants of hard and soft margin optimization problems over p+n instances. The resulting classifiers of our methods are guaranteed to have a certain amount of AUC scores.

元の言語英語
ページ(範囲)719-729
ページ数11
ジャーナルIEICE Transactions on Information and Systems
E101D
発行部数3
DOI
出版物ステータス出版済み - 3 1 2018

Fingerprint

Classifiers

All Science Journal Classification (ASJC) codes

  • Software
  • Hardware and Architecture
  • Computer Vision and Pattern Recognition
  • Electrical and Electronic Engineering
  • Artificial Intelligence

これを引用

Efficient reformulation of 1-norm ranking SVM. / Suehiro, Daiki; Hatano, Kohei; Takimoto, Eiji.

:: IEICE Transactions on Information and Systems, 巻 E101D, 番号 3, 01.03.2018, p. 719-729.

研究成果: ジャーナルへの寄稿記事

@article{d52b6e9cecd04cbaaa922848917969d7,
title = "Efficient reformulation of 1-norm ranking SVM",
abstract = "Finding linear functions that maximize AUC scores is important in ranking research. A typical approach to the ranking problem is to reduce it to a binary classification problem over a new instance space, consisting of all pairs of positive and negative instances. Specifically, this approach is formulated as hard or soft margin optimization problems over pn pairs of p positive and n negative instances. Solving the optimization problems directly is impractical since we have to deal with a sample of size pn, which is quadratically larger than the original sample size p + n. In this paper, we reformulate the ranking problem as variants of hard and soft margin optimization problems over p+n instances. The resulting classifiers of our methods are guaranteed to have a certain amount of AUC scores.",
author = "Daiki Suehiro and Kohei Hatano and Eiji Takimoto",
year = "2018",
month = "3",
day = "1",
doi = "10.1587/transinf.2017EDP7233",
language = "English",
volume = "E101D",
pages = "719--729",
journal = "IEICE Transactions on Information and Systems",
issn = "0916-8532",
publisher = "一般社団法人電子情報通信学会",
number = "3",

}

TY - JOUR

T1 - Efficient reformulation of 1-norm ranking SVM

AU - Suehiro, Daiki

AU - Hatano, Kohei

AU - Takimoto, Eiji

PY - 2018/3/1

Y1 - 2018/3/1

N2 - Finding linear functions that maximize AUC scores is important in ranking research. A typical approach to the ranking problem is to reduce it to a binary classification problem over a new instance space, consisting of all pairs of positive and negative instances. Specifically, this approach is formulated as hard or soft margin optimization problems over pn pairs of p positive and n negative instances. Solving the optimization problems directly is impractical since we have to deal with a sample of size pn, which is quadratically larger than the original sample size p + n. In this paper, we reformulate the ranking problem as variants of hard and soft margin optimization problems over p+n instances. The resulting classifiers of our methods are guaranteed to have a certain amount of AUC scores.

AB - Finding linear functions that maximize AUC scores is important in ranking research. A typical approach to the ranking problem is to reduce it to a binary classification problem over a new instance space, consisting of all pairs of positive and negative instances. Specifically, this approach is formulated as hard or soft margin optimization problems over pn pairs of p positive and n negative instances. Solving the optimization problems directly is impractical since we have to deal with a sample of size pn, which is quadratically larger than the original sample size p + n. In this paper, we reformulate the ranking problem as variants of hard and soft margin optimization problems over p+n instances. The resulting classifiers of our methods are guaranteed to have a certain amount of AUC scores.

UR - http://www.scopus.com/inward/record.url?scp=85042679921&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85042679921&partnerID=8YFLogxK

U2 - 10.1587/transinf.2017EDP7233

DO - 10.1587/transinf.2017EDP7233

M3 - Article

AN - SCOPUS:85042679921

VL - E101D

SP - 719

EP - 729

JO - IEICE Transactions on Information and Systems

JF - IEICE Transactions on Information and Systems

SN - 0916-8532

IS - 3

ER -