VLSI IMPLEMENTATION OF LOCALLY CONNECTED NEURAL-NETWORK FOR SOLVING PARTIAL-DIFFERENTIAL EQUATIONS

This brief presents a locally connected neural network for solving a c lass of partial differential equations. Each neural fell is designed u sing active and passive components. An architecture is described to co ntrol the weights between the neurons, The major benefit of the archit ecture is that it does not require additional space outside of the cel l for routing the control lines no matter how many cells are used, The CMOS VLSI implementation of a sixteen cell network was fabricated and measured. The results of this network are compared to the numerical s olution of the partial differential equations.