AN IMPROVEMENT OF THE PSEUDOINVERSE RULE WITH DIAGONAL ELEMENTS

A pseudoinverse rule, one of major rule to determine a weight matrix f or associative memory, has large capacity comparing with other determi ning rules. However, it is well-known that the rule has small domains of attraction of memory vectors on account of many spurious states. In this paper, we try to improve the problem by means of subtracting a c onstant from all diagonal elements of a weight matrix. By this method, many spurious states disappear and eigenvectors with negative eigenva lues are introduced for the orthocomplement of the subspace spanned by memory vectors. This method can be applied to two types of networks: binary network and analog network. Some computer simulations are perfo rmed for both two models. The results of the simulations show our impr ovement is effective to extend error correcting ability for both netwo rks.