Classification Based on Homogeneous Logical Proportions

Publisher:
Springer
Publication Type:
Conference Proceeding
Citation:
SGAI Conf., 2013, pp. 53 - 60
Issue Date:
2013
Full metadata record
Files in This Item:
Filename Description Size
10.1007%2F978-3-319-02621-3_4.pdfPublished version154.17 kB
Adobe PDF
A classification method based on a measure of analogical dissimilarity has been proposed some years ago by Laurent Miclet and his colleagues, which was giving very good results. We restart this study on a slightly different basis. Instead of estimating analogical dissimilarity, we use the logical definition of an analogical proportion. We also consider other related logical proportions and their link with analogical proportion. The paper reports on an ongoing work and contributes to a comparative study of the logical proportions predictive accuracy on a set of standard benchmarks coming from UCI repository. Logical proportions constitute an interesting framework to deal with binary and/or nominal classification tasks without introducing any metrics or numerical weights.
Please use this identifier to cite or link to this item: