A memristive deep belief neural network based on silicon synapses

TL;DR

This paper presents silicon-based memristive synapses integrated in CMOS technology, enabling efficient deep belief neural networks with high accuracy and energy efficiency for pattern recognition tasks.

Contribution

It introduces floating gate memristive devices fabricated in CMOS, demonstrating their suitability for neuromorphic computing and deep belief networks.

Findings

Achieved 97.05% accuracy on MNIST dataset.

Provided two orders of magnitude higher energy efficiency than GPUs.

Demonstrated in-situ training of memristive neural networks.

Abstract

Memristor-based neuromorphic computing could overcome the limitations of traditional von Neumann computing architectures -- in which data are shuffled between separate memory and processing units -- and improve the performance of deep neural networks. However, this will require accurate synaptic-like device performance, and memristors typically suffer from poor yield and a limited number of reliable conductance states. Here we report floating gate memristive synaptic devices that are fabricated in a commercial complementary metal-oxide-semiconductor (CMOS) process. These silicon synapses offer analogue tunability, high endurance, long retention times, predictable cycling degradation, moderate device-to-device variations, and high yield. They also provide two orders of magnitude higher energy efficiency for multiply-accumulate operations than graphics processing units. We use two 12-by-8 arrays of the memristive devices for in-situ training of a 19-by-8 memristive restricted Boltzmann machine for pattern recognition via a gradient descent algorithm based on contrastive divergence. We then create a memristive deep belief neural network consisting of three memristive restricted Boltzmann machines. We test this on the modified National Institute of Standards and Technology (MNIST) dataset, demonstrating recognition accuracy up to 97.05%.