A Hopfield neural network solution to the TCM partitioning problem

In this paper, the problem of circuit partitioning for multichip module packaging realization is considered. A Hopfield-type neural network is designed for use in solving the previously proposed thermal conduction module (TCM) partitioning problem, to which only heuristic solutions have been applied thus far. It is expected that the global nature of the Hopfield neural network's operation will make it a useful tool as the problem complexity is increased.<
>