It is shown that ART is partially parallel processed by systolic algorithm with matrix-vector multiplication in the learning phase. In the search phase, also MAXNET is adopted because of systolic algorithm. Without limitation of neuron number, this al...
It is shown that ART is partially parallel processed by systolic algorithm with matrix-vector multiplication in the learning phase. In the search phase, also MAXNET is adopted because of systolic algorithm. Without limitation of neuron number, this algorithm is designed. Since computation of weights sum is parallelly processed with full pipelinability, time complexity of O(n^(2)) is reduced to O(n).