VLSI NEURAL-NETWORK CIRCUIT USING A MULTIPLE-INPUT TRANSCONDUCTANCE AMPLIFIER AND DIGITAL MULTIPLEXERS

A neural network is designed using a multiple-input transconductance a mplifier (MITA) and digital multiplexers. The MITA is used to construc t the summation of synaptic weights and digital multiplexers are used to provide a more precise multiplication operation and a smaller circu it size. Short-term memory activities of the neural network obtained f rom Pspice circuit simulation compare favourably with those calculated directly from the neural network algorithm.