Constructive neural-network learning algorithms for pattern classification

Constructive learning algorithms offer an attractive approach for the incre mental construction of near-minimal neural-network architectures for patter n classification. They help overcome the need for ad hoc and often inapprop riate choices of network topology in algorithms that search for suitable we ights in a priori fixed network architectures. Several such algorithms are proposed in the literature and shown to converge to zero classification err ors (under certain assumptions) on tasks that involve learning a binary to binary mapping (i.e., classification problems involving binary-valued input attributes and two output categories), We present two constructive learnin g algorithms MPyramid-real and MTiling-real that extend the pyramid and til ing algorithms, respectively, for learning real to M-ary mappings (i.e., cl assification problems involving real-valued input attributes and multiple o utput classes). Ne prove the convergence of these algorithms and empiricall y demonstrate their applicability to practical pattern classification probl ems. Additionally, we show how the incorporation of a local pruning step ca n eliminate several redundant neurons from MTiling-real networks.