Energy efficient programming of nanoelectronic synaptic devices for large-scale implementation of associative and temporal sequence learning

Abstract

A nanoscale, two-terminal device emulating plasticity and energy efficiency of biological synapses is a critical element for realizing brain-inspired computational systems and real-time brain simulators. In this work, we explore the use of phase change materials (PCM), widely used for memory applications, to build electronic synapses which implement synaptic plasticity with picojoule level energy consumption. Gradual switching characteristics and different spike schemes are discussed from implementation of synaptic plasticity and energy consumption perspectives. Our simulations demonstrate that a recurrent network of PCM synapses in a crossbar array can achieve brain-like associative and temporal sequence learning. Asymmetric plasticity is shown to transform temporal information into spatial information for sequence learning. Symmetric plasticity enables the storage and recall of certain patterns associatively by acting as a coincidence detector for neuronal activity.