Neuromorphic computing with multi-memristive synapses

TLDR

Neuromorphic computing promises next‑generation intelligent systems, and memristive devices can encode synaptic weights, but achieving precise, wide‑range conductance modulation remains difficult. The authors propose a multi‑memristive synaptic architecture that uses a global counter‑based arbitration scheme to overcome this conductance‑modulation challenge. They model phase‑change memory devices, simulate the architecture’s performance on both spiking and non‑spiking neural networks, and validate the design through extensive simulations. Experimental tests on over a million phase‑change memory devices demonstrate unsupervised learning of temporal correlations in a spiking network, marking a significant advance toward large‑scale, energy‑efficient neuromorphic systems.

Abstract

Neuromorphic computing has emerged as a promising avenue towards building the next generation of intelligent computing systems. It has been proposed that memristive devices, which exhibit history-dependent conductivity modulation, could efficiently represent the synaptic weights in artificial neural networks. However, precise modulation of the device conductance over a wide dynamic range, necessary to maintain high network accuracy, is proving to be challenging. To address this, we present a multi-memristive synaptic architecture with an efficient global counter-based arbitration scheme. We focus on phase change memory devices, develop a comprehensive model and demonstrate via simulations the effectiveness of the concept for both spiking and non-spiking neural networks. Moreover, we present experimental results involving over a million phase change memory devices for unsupervised learning of temporal correlations using a spiking neural network. The work presents a significant step towards the realization of large-scale and energy-efficient neuromorphic computing systems.