Data Exploration and Linear Separability

This book presents a unified framework for pattern analysis based on minimization of convex and piecewise linear (CPL) criterion functions. These criterion functions can be linked to the Perceptron theory of pattern recognition and learning algorithms of formal neurons. The basis exchange algorithms provide effective method of CPL functions minimization. From this unifying perspective and for the first time collected in one book the theoretical foundation of classification regression clustering and visualization with some applications to the analysis of medical data are discussed. The book is intended for the data mining community concerned with computational foundations and the design of data analysis systems.