Output details

Delayed switching applied to memristor neural networks

<01> One difficulty in building a computer that works in a way similar to the human brain is that it takes about the same silicon area to emulate a CMOS synapse as is needed to emulate a neuron. On average, each neuron is connected to other neurons through about 20,000 synapses. This paper demonstrates that the memristor provides a new way of implementing synapses, which narrows significantly the “four orders of magnitude” gap in silicon area required. This work is based on the “delayed switching effect” described in Output 1, and has received interest from the Learning Systems community.