Special Issue of Interface on Neuromorphic Computing: An Introduction and State of the Field

Abstract

The human brain integrates and processes information to perform complex cognitive tasks within approximately 20 watts of power. Today’s fastest supercomputer is unable to deliver the power requirements and the number of operations at the same energy levels. In the brain, the discrete and sparse events in time called spikes are used to process and encode the information. The energy efficiency of the brain is attributed to the sparsity of the spikes and event-driven communication between the neurons. Complex interconnections among the 1011 neurons and 1015 synapses in the human brain process the information, possibly encoded in the time, frequency, and phase of the spikes. Therefore, to emulate human cognition requires novel electronic devices and new algorithmic approaches. Brain-inspired computing, or neuromorphic computing, is an approach to build energy-efficient computing architectures and systems.