Neuromorphic computing using non-volatile memory

TLDR

Dense crossbar arrays of non‑volatile memory devices enable massively‑parallel, energy‑efficient neuromorphic computing. The paper reviews recent advances and surveys research on using various NVM devices as synapses or neurons in spiking neural networks, deep neural networks, and memcomputing. The authors describe how NVM synapses implement spike‑timing‑dependent plasticity in spiking neural networks, perform analog matrix–vector multiplication for deep neural networks, and evaluate device properties such as conductance range, linearity, retention, endurance, switching power, and variability. The authors suggest that NVM‑based neuromorphic computing could markedly improve power efficiency and speed over GPU‑based deep neural network training for commercial applications.

Abstract

Dense crossbar arrays of non-volatile memory (NVM) devices represent one possible path for implementing massively-parallel and highly energy-efficient neuromorphic computing systems. We first review recent advances in the application of NVM devices to three computing paradigms: spiking neural networks (SNNs), deep neural networks (DNNs), and 'Memcomputing'. In SNNs, NVM synaptic connections are updated by a local learning rule such as spike-timing-dependent-plasticity, a computational approach directly inspired by biology. For DNNs, NVM arrays can represent matrices of synaptic weights, implementing the matrix–vector multiplication needed for algorithms such as backpropagation in an analog yet massively-parallel fashion. This approach could provide significant improvements in power and speed compared to GPU-based DNN training, for applications of commercial significance. We then survey recent research in which different types of NVM devices – including phase change memory, conductive-bridging RAM, filamentary and non-filamentary RRAM, and other NVMs – have been proposed, either as a synapse or as a neuron, for use within a neuromorphic computing application. The relevant virtues and limitations of these devices are assessed, in terms of properties such as conductance dynamic range, (non)linearity and (a)symmetry of conductance response, retention, endurance, required switching power, and device variability.