Kernel partial least squares regression in reproducing kernel hilbert space
The Journal of Machine Learning Research
IEEE/ACM Transactions on Computational Biology and Bioinformatics (TCBB)
Gene selection from microarray data for cancer classification-a machine learning approach
Computational Biology and Chemistry
Learning kernel subspace classifier
ICB'07 Proceedings of the 2007 international conference on Advances in Biometrics
| Hi-index | 0.00 |
The classification of cancer based on gene expression data is one of the most important tasks in bioinformatics, and is essential for future clinical implementations of microarray based cancer diagnosis. In this paper, a novel procedure for classifying cancer using the gene expression data is proposed based on a Kernel Principal Component Self-regression (KPCSR) model. Developed from Kernel Principal Component Analysis (KPCA), the KPCSR model selects a subset of the principal components from the kernel space for the input variables to regress in order to accurately characterize each type of cancer. A modular scheme with class-specific KPCSR structure proves very efficient, from which each cancer class is assigned an independent KPCSR model for coding the corresponding gene expression information. The performance was measured on several public gene expression datasets involving human tumor samples, using 5-fold cross-validation and leave-one-out cross-validation (LOOCV) respectively. Experimental results has shown that the classification accuracies are better or comparable to the maximum accuracies based on the Support Vector Machine and k-Nearest Neighbor classifications combined with various gene selection schemes reported previously in the literature. These results suggest that our proposed method is useful for microarray based cancer classification.