Statistical and Machine-Learning Methods for Class Prediction in High Dimension

In recent years, the rapid progress in molecular biology and biotechnology has enabled widespread and increasing use of high-throughput technologies. These technologies, which exhaustively monitor the characteristics of biological specimens at the molecular level, include genome-wide singlenucleotide polymorphism typing, gene-expression microarrays, and protein mass spectrometry. One mission for statistics and bioinformatics is to develop novel data-learning methodologies for pattern recognition and prediction or classiˆcation through extracting useful information from these

CONTENTS

14.1 Introduction ................................................................................................253 14.2 Framework of Classiˆcation .....................................................................254