Neuromorphic computing using non-volatile memory

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 nonfilamentary 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.

Bridging Biological and Artificial Neural Networks with Emerging Neuromorphic Devices: Fundamentals, Progress, and Challenges

“…[72][73][74] Here we pick some representative implementations as listed in Table 1. [72][73][74] Here we pick some representative implementations as listed in Table 1.…”

A Robust Artificial Synapse Based on Organic Ferroelectric Polymer

“…The most-reported memristors focus on filament-forming metal oxides (FFMOs) [5] and phase change memory (PCM) materials. [2] To improve homogeneity of artificial neural network, purely electronic memristors based on ferroelectrics are proposed to emulate neuronal and synaptic functions. [2,5b] This is particularly problematic for neuromorphic applications, since a single highly conductive device will contribute much more current into a vector sum than its neighbors.…”

Mechanisms for Enhanced State Retention and Stability in Redox‐Gated Organic Neuromorphic Devices

“…[8][9][10] Among these, resistive memory (memristive) devices [11,12] have gained traction as a highly suitable option, offering projected efficiency gains of up to 10 6 over von Neumann architectures when implementing ANN algorithms. [16][17][18] By arranging these devices in a crossbar array architecture, vector-matrix multiplication (VMM) can be performed in an analog fashion where the input vector is represented by voltages, the operator matrix is represented by the conductance of each memristive element, and the output vector is represented by currents. [16][17][18] By arranging these devices in a crossbar array architecture, vector-matrix multiplication (VMM) can be performed in an analog fashion where the input vector is represented by voltages, the operator matrix is represented by the conductance of each memristive element, and the output vector is represented by currents.…”