Kernel associative memory (KAM) has previously been proposed as an efficient scheme for face recognition. In this paper, a hybrid method of combining KAM and Gabor wavelet transfo...
—This paper studies the problem of understanding noisy and structurally deformed two-dimensional images by means of abstractly defined neural works. First, in the framework of sy...
—In this paper, a new model that can ultimately create its own set of perceptual features is proposed. Using a bidirectional associative memory (BAM)-inspired architecture, the r...
This paper investigates possible connection strategies in sparsely connected associative memory models. This is interesting because real neural networks must have both efficient p...
Abstract. Our work is concerned with finding optimum connection strategies in highperformance associative memory models. Taking inspiration from axonal branching in biological neur...