Application of Statistical Learning Theory to DNA Microarray Analysis by

This thesis focuses on applying Support Vector Machines SVMs an algorithm founded in the framework of statistical learning theory to analyzing DNA microarray data The rst part of the thesis focuses on extensions to SVMs required for analyzing microarray data First the problem of choosing multiple parameters at once for SVMs is addressed This is used as the basis of a feature selection algorithm that allows us to select which genes are most relevant in discriminating between two classes A methodology for outputting con dence levels as well as class labels is developed The second part of the thesis consists of a systematic evaluation of a variety of machine learning algorithms on ve datasets from four types of molecular cancer classi cation problems It will also describe some very promising results in predicting treatment outcome from expression data for brain tumors and lymphoma The al gorithms compared will be k Nearest Neighbors kNN Naive Bayes NB Weighted Voting Average WV and Support Vector Machines SVMs Learning curves are constructed for the lymphoma treatment and morphology datasets to compare per formance as a function of sample size and try to address the questions of given enough data can error rates that are clinically acceptable be achieved and how much data is needed to achieve such a rate A simple analytic model is constructed to estimate the variance in classi cation accuracy due to sample size limitations Thesis Supervisor Tomaso Poggio Title Uncas and Helen Whitaker Professor of Brain Sciences