Testing Exchangeability for Transfer Decision

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

This paper introduces a non-parametric test to decide whether to transfer data from a source domain to a target domain to improve the generalization performance of predictive models on the target domain. The test is based on the conformal prediction framework: it statistically tests whether the target and source data are generated from the same distribution under the exchangeability assumption. The experiments show that the test is capable of outperforming existing methods when it decides on instance transfer. 

Original languageEnglish
Pages (from-to)64-71
Number of pages8
JournalPattern Recognition Letters
Volume88
DOIs
Publication statusPublished - 1 Mar 2017

Keywords

  • Conformity prediction framework
  • Exchangeability test
  • Instance-transfer learning
  • SELECTION

Cite this

@article{29633a8553cc4b6ab6b72d867e62f680,
title = "Testing Exchangeability for Transfer Decision",
abstract = "This paper introduces a non-parametric test to decide whether to transfer data from a source domain to a target domain to improve the generalization performance of predictive models on the target domain. The test is based on the conformal prediction framework: it statistically tests whether the target and source data are generated from the same distribution under the exchangeability assumption. The experiments show that the test is capable of outperforming existing methods when it decides on instance transfer. ",
keywords = "Conformity prediction framework, Exchangeability test, Instance-transfer learning, SELECTION",
author = "Shuang Zhou and Evgueni Smirnov and Gijsbertus Schoenmakers and Kurt Driessens and Ralf Peeters",
year = "2017",
month = "3",
day = "1",
doi = "10.1016/j.patrec.2016.12.021",
language = "English",
volume = "88",
pages = "64--71",
journal = "Pattern Recognition Letters",
issn = "0167-8655",
publisher = "Elsevier",

}

Testing Exchangeability for Transfer Decision. / Zhou, Shuang; Smirnov, Evgueni; Schoenmakers, Gijsbertus; Driessens, Kurt; Peeters, Ralf.

In: Pattern Recognition Letters, Vol. 88, 01.03.2017, p. 64-71.

Research output: Contribution to journalArticleAcademicpeer-review

TY - JOUR

T1 - Testing Exchangeability for Transfer Decision

AU - Zhou, Shuang

AU - Smirnov, Evgueni

AU - Schoenmakers, Gijsbertus

AU - Driessens, Kurt

AU - Peeters, Ralf

PY - 2017/3/1

Y1 - 2017/3/1

N2 - This paper introduces a non-parametric test to decide whether to transfer data from a source domain to a target domain to improve the generalization performance of predictive models on the target domain. The test is based on the conformal prediction framework: it statistically tests whether the target and source data are generated from the same distribution under the exchangeability assumption. The experiments show that the test is capable of outperforming existing methods when it decides on instance transfer. 

AB - This paper introduces a non-parametric test to decide whether to transfer data from a source domain to a target domain to improve the generalization performance of predictive models on the target domain. The test is based on the conformal prediction framework: it statistically tests whether the target and source data are generated from the same distribution under the exchangeability assumption. The experiments show that the test is capable of outperforming existing methods when it decides on instance transfer. 

KW - Conformity prediction framework

KW - Exchangeability test

KW - Instance-transfer learning

KW - SELECTION

U2 - 10.1016/j.patrec.2016.12.021

DO - 10.1016/j.patrec.2016.12.021

M3 - Article

VL - 88

SP - 64

EP - 71

JO - Pattern Recognition Letters

JF - Pattern Recognition Letters

SN - 0167-8655

ER -