The Generalization of a Constructive Algorithm in Pattern Classification Problems

The use of a constructive algorithm for pattern classi cation is examined. The algorithm, a `Perceptron Cascade', has been shown to converge to zero errors whilst learning any consistent classi cation of real-valued pattern vectors (Burgess, 1992). Limiting network size and producing bounded decision regions are noted to be important for the generalization ability of a network. A scheme is suggested by which a result on generalization (Vapnik, 1992) may enable calculation of the optimal network size. A fast algorithm for principal component analysis (Sirat, 1991) is used to construct `hyper-boxes' around each class of patterns to ensure bounded decision regions. Performance is compared with the Gaussian Maximum Likelihood procedure in three arti cial problems simulating real pattern classi cation applications.