Associative Learning Using Ising-Like Model

In this paper, a new computational model of associative learning is proposed, which is based on the Ising model. Application of the stochastic gradient descent algorithm to the proposed model yields an on-line learning rule. Next, it is shown that the obtained new learning rule generalizes two well-known learning rules, i.e., the Hebbian rule and the Oja's rule. Later, the fashion of incorporating the cognitive account into the obtained associative learning rule is proposed. At the end of the paper, experiments were carried out for testing the backward blocking and the reduced overshadowing and blocking phenomena. The obtained results are discussed and conclusions are drawn.