Classification with no discrimination by preferential sampling

F. Kamiran, T.G.K. Calders

    Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

    1 Downloads (Pure)


    The concept of classification without discrimination is a new area of research. (Kamiran & Calders, 2009) introduced the idea of Classification with No Discrimination (CND) and proposed a solution based on "massaging" the data to remove the discrimination from it with the least possible changes. In this paper, we propose a new solution to the CND problem by introducing a sampling scheme for making the data discrimination free instead of relabeling the dataset. On the resulting non-discriminatory dataset we then learn a classifier. This new method is not only less intrusive as compared to the "massaging" but also outperforms the "reweighing" approach of (Calders et al., 2009). The proposed method has been implemented and experimental results on the Census Income dataset show promising results: in all experiments our method performs onpar with the state-of-the art non-discriminatory techniques.
    Original languageEnglish
    Title of host publicationInformal proceedings of the 19th Annual Machine Learning Conference of Belgium and The Netherlands (Benelearn'10, Leuven, Belgium, May 27-28, 2010)
    Publication statusPublished - 2010


    Dive into the research topics of 'Classification with no discrimination by preferential sampling'. Together they form a unique fingerprint.

    Cite this