A systolic array implementation of a dynamic sequential neural network for pattern recognition

The authors have previously developed a sequential algorithm for training a multi-layer perceptron classifier (1993). The idea is to exploit the fact that the locations of boundary segments are local divisions. Training is achieved by updating local covariances using the recursive least squares (RLS) algorithm. The algorithm is sequential in the sense that training examples are passed only once, and the network will learn and/or expand at the arrival of each example. The major advantage in this sequential scheme is the feasibility of pipelining the training procedures in a true parallel architecture. The authors present a systolic array implementation of the sequential input space partitioning (SISP) algorithm