Regularization paths of L1-penalized ROC Curve-Optimizing Support Vector Machines

Hyungwoo Kim; Insuk Sohn; Seung Jun Shin

doi:10.1002/sta4.400

Regularization paths of L₁-penalized ROC Curve-Optimizing Support Vector Machines

Hyungwoo Kim
, Insuk Sohn
, Seung Jun Shin^*

^*Corresponding author for this work

Research output: Contribution to journal › Article › peer-review

2 Citations (Scopus)

Abstract

The receiver operator characteristic (ROC) curve is one of the most popular tools to evaluate the performance of binary classifiers in a variety of applications. Rakotomamonjy (2004) proposed the ROC-SVM that directly optimizes the area under the ROC curve instead of the prediction accuracy. In this article, we study the L₁-penalized ROC-SVM that directly optimizes the ROC curve. We first show that the L₁-penalized ROC-SVM has piecewise linear regularization paths and then develop an efficient algorithm to compute the entire paths, which greatly facilitates its tuning procedure.

Original language	English
Article number	e400
Journal	Stat
Volume	10
Issue number	1
DOIs	https://doi.org/10.1002/sta4.400
Publication status	Published - 2021 Dec

Bibliographical note

Publisher Copyright:
© 2021 John Wiley & Sons, Ltd.

ASJC Scopus subject areas

Statistics and Probability
Statistics, Probability and Uncertainty

Access to Document

10.1002/sta4.400

Cite this

@article{e590d981393847b1b67ec24839ce453d,

title = "Regularization paths of L1-penalized ROC Curve-Optimizing Support Vector Machines",

abstract = "The receiver operator characteristic (ROC) curve is one of the most popular tools to evaluate the performance of binary classifiers in a variety of applications. Rakotomamonjy (2004) proposed the ROC-SVM that directly optimizes the area under the ROC curve instead of the prediction accuracy. In this article, we study the L1-penalized ROC-SVM that directly optimizes the ROC curve. We first show that the L1-penalized ROC-SVM has piecewise linear regularization paths and then develop an efficient algorithm to compute the entire paths, which greatly facilitates its tuning procedure.",

author = "Hyungwoo Kim and Insuk Sohn and Shin, \{Seung Jun\}",

note = "Publisher Copyright: {\textcopyright} 2021 John Wiley \& Sons, Ltd.",

year = "2021",

month = dec,

doi = "10.1002/sta4.400",

language = "English",

volume = "10",

journal = "Stat",

issn = "2049-1573",

publisher = "John Wiley and Sons Inc.",

number = "1",

}

TY - JOUR

T1 - Regularization paths of L1-penalized ROC Curve-Optimizing Support Vector Machines

AU - Kim, Hyungwoo

AU - Sohn, Insuk

AU - Shin, Seung Jun

PY - 2021/12

Y1 - 2021/12

N2 - The receiver operator characteristic (ROC) curve is one of the most popular tools to evaluate the performance of binary classifiers in a variety of applications. Rakotomamonjy (2004) proposed the ROC-SVM that directly optimizes the area under the ROC curve instead of the prediction accuracy. In this article, we study the L1-penalized ROC-SVM that directly optimizes the ROC curve. We first show that the L1-penalized ROC-SVM has piecewise linear regularization paths and then develop an efficient algorithm to compute the entire paths, which greatly facilitates its tuning procedure.

AB - The receiver operator characteristic (ROC) curve is one of the most popular tools to evaluate the performance of binary classifiers in a variety of applications. Rakotomamonjy (2004) proposed the ROC-SVM that directly optimizes the area under the ROC curve instead of the prediction accuracy. In this article, we study the L1-penalized ROC-SVM that directly optimizes the ROC curve. We first show that the L1-penalized ROC-SVM has piecewise linear regularization paths and then develop an efficient algorithm to compute the entire paths, which greatly facilitates its tuning procedure.

UR - https://www.scopus.com/pages/publications/85121783368

U2 - 10.1002/sta4.400

DO - 10.1002/sta4.400

M3 - Article

AN - SCOPUS:85121783368

SN - 2049-1573

VL - 10

JO - Stat

JF - Stat

IS - 1

M1 - e400

ER -