Classification of Proteomic Data with Logistic Kernel Partial Least Squares Algorithm

In this paper we introduce the logistic kernel partial least squares (LKPLS) algorithm for classi?cation of health vs. cancer using mass spectrometry (MS). Wavelet decomposition is proposed for feature selection and data preprocessing. LKPLS combines the logistic regression with the kernel partial least squares algorithm. The method is applied to real life cancer samples. Experimental comparisons show that LKPLS outperforms other methods in the analysis of MS data.