SOT-MRAM-Enabled Probabilistic Binary Neural Networks for Noise-Tolerant and Fast Training

TL;DR

This paper introduces a novel SOT-MRAM-based probabilistic binary neural network that offers noise-tolerance, fast training, and high accuracy, demonstrating significant improvements over traditional neural network implementations.

Contribution

The work presents a new SOT-MRAM device for probabilistic neural networks, enabling fast, noise-tolerant training and high on-chip inference performance.

Findings

Achieved 18x faster training speed compared to traditional neural networks.

Maintained over 97% accuracy under noise perturbations.

Demonstrated near-ideal MNIST inference performance on-chip.

Abstract

We report the use of spin-orbit torque (SOT) magnetoresistive random-access memory (MRAM) to implement a probabilistic binary neural network (PBNN) for resource-saving applications. The in-plane magnetized SOT (i-SOT) MRAM not only enables field-free magnetization switching with high endurance (> 10^11), but also hosts multiple stable probabilistic states with a low device-to-device variation (< 6.35%). Accordingly, the proposed PBNN outperforms other neural networks by achieving an 18* increase in training speed, while maintaining an accuracy above 97% under the write and read noise perturbations. Furthermore, by applying the binarization process with an additional SOT-MRAM dummy module, we demonstrate an on-chip MNIST inference performance close to the ideal baseline using our SOT-PBNN hardware.