Large machine-part family formation utilizing a parallel ART1 neural network

The binary adaptive resonance (ART1) neural network algorithm has been succ essfully implemented in the past for the classifying and grouping of simila r vectors from a machine-part matrix. A modified ART1 paradigm which reorde rs the input vectors, along with a modified procedure for storing a group's representation vectors, has proven successful in both speed and functional ity in comparison to former techniques. This paradigm has been adapted and implemented on a neuro-computer utilizing 256 processors which allows the c omputer to take advantage of the inherent parallelism of the ART1 algorithm . The parallel implementation results in tremendous improvements in the spe ed of the machine-part matrix optimization. The machine-part matrix was ini tially limited to 65,536 elements (256x256) which is a consequence of the m aximum number of processors within the parallel computer. The restructuring and modification of the parallel implementation has allowed the number of matrix elements to increase well beyond their previous limits. Comparisons of the modified structure with both the serial algorithm and the initial pa rallel implementation are made. The advantages of using a neural network ap proach in this case are discussed.