Pattern Recognition on Oriented Matroids

Committees for Pattern Recognition: Infeasible Systems of Linear Inequalities, Hyperplane Arrangements, and Realizable Oriented Matroids

In supervised learning to recognize, through the synthesis of decision rules that use separating surfaces in the feature space, a recognition complex (whose concept is relevant to the mathematical constructs discussed in this book) is provided with the training patterns contained in a training set, together with class labels which are associated with the patterns. The training patterns from the same class compose a training sample. Thus, in the two-class pattern recognition problem, the training set is a disjoint union of two training samples. The recognition complex classifies new patterns as representatives ...