Classication of Breast Cancer Histopathological Images using Adaptive Penalized Logistic Regression with Wilcoxon Rank Sum Test

Classication of the histopathological image is an important problem indiagnosis and treatment. The problem of selecting the most useful fea-tures from thousands of candidates is a key problem in classication of thehistopathological image. In this paper, an adaptive penalized logistic regres-sion is proposed, with the aim of identication features, by combining thelogistic regression with the weighted L1-norm. Our proposed method is ex-perimentally tested and compared with state-of-the-art methods based on apublicly recent breast cancer histopathological image datasets. The resultsshow that the proposed method signicantly outperforms three competitormethods in terms of overall classication accuracy and the number of selectedfeatures.