Pattern Recognition, 4th Edition

Chapter 9. Context-Dependent Classification

9.1. Introduction

The classification tasks considered so far have assumed that no relation exists among the various classes. In other words, having obtained a feature vector x from a class ωⁱ, the next feature vector could belong to any other class. In this chapter we will remove this assumption, and we will assume that the various classes are closely related. That is, successive feature vectors are not independent. Under such an assumption, classifying each feature vector separately from the others obviously has no meaning. The class to which a feature vector is assigned depends (a) on its own value, (b) on the values of the other feature vectors, and (c) on the existing relation among the various ...