90
4

Learning by random walks in the weight space of the Ising perceptron

Abstract

Several variants of a stochastic local search process for constructing the synaptic weights of an Ising perceptron are studied. In this process, binary patterns are sequentially presented to the Ising perceptron and are then learned as the synaptic weight configuration is modified through a chain of single- or double-weight flips within the compatible weight configuration space of the earlier learned patterns. This process is able to reach a storage capacity of α0.63\alpha \approx 0.63 for pattern length N = 101 and α0.41\alpha \approx 0.41 for N = 1001. If in addition a relearning process is exploited, the learning performance is further improved to a storage capacity of α0.80\alpha \approx 0.80 for N = 101 and α0.42\alpha \approx 0.42 for N=1001. We found that, for a given learning task, the solutions constructed by the random walk learning process are separated by a typical Hamming distance, which decreases with the constraint density α\alpha of the learning task; at a fixed value of α\alpha, the width of the Hamming distance distributions decreases with NN.

View on arXiv
Comments on this paper