A pulse stream system for low-power neuro-fuzzy computation

This paper describes a VLSI device designed for low-power Neuro-Fuzzy computation, which is based on Coherent Pulse Width Modulation. The device implements several Neural and Fuzzy paradigms. Weights are stored as a voltage on a pair of capacitors, which are sequentially refreshed by a built-in self-refresh circuit. The performance of the system is analyzed theoretically, and results are compared against values measured from a prototype device, which contains a 32/spl times/32 synaptic array, consuming 10 mW of power at 140 MCPS.