Combination of One-Class Support Vector Machines for Classification with Reject Option

Blaise Hanczar 1, 2, * Michèle Sebag 3, 2
* Corresponding author
2 TAO - Machine Learning and Optimisation
LRI - Laboratoire de Recherche en Informatique, UP11 - Université Paris-Sud - Paris 11, Inria Saclay - Ile de France, CNRS - Centre National de la Recherche Scientifique : UMR8623
Abstract : This paper focuses on binary classification with reject op-tion, enabling the classifier to detect and abstain hazardous decisions. While reject classification produces in more reliable decisions, there is a tradeoff between accuracy and rejection rate. Two type of rejection are considered: ambiguity and outlier rejection. The state of the art mostly handles ambiguity rejection and ignored outlier rejection. The proposed approach, referred as CONSUM, handles both ambiguity and outliers detection. Our method is based on a quadratic constrained optimization formulation, combining one-class support vector machines. An adapta-tion of the sequential minimal optimization algorithm is proposed to solve the minimization problem. The experimental study on both artifi-cial and real world datasets exams the sensitivity of the CONSUM with respect to the hyper-parameters and demonstrates the superiority of our approach.
Document type :
Conference papers
Complete list of metadatas

Cited literature [15 references]  Display  Hide  Download

https://hal.inria.fr/hal-01109774
Contributor : Marc Schoenauer <>
Submitted on : Thursday, January 29, 2015 - 2:04:59 PM
Last modification on : Monday, December 9, 2019 - 5:24:06 PM
Long-term archiving on: Saturday, April 15, 2017 - 11:57:50 PM

File

hanczarSebagECML2014-author.pd...
Files produced by the author(s)

Identifiers

Collections

Citation

Blaise Hanczar, Michèle Sebag. Combination of One-Class Support Vector Machines for Classification with Reject Option. Machine Learning and Knowledge Discovery in Databases - Part I, Sep 2014, Nancy, France. pp.547 - 562, ⟨10.1007/978-3-662-44848-9_35⟩. ⟨hal-01109774⟩

Share

Metrics

Record views

551

Files downloads

744