Convolutional Neural Networks Utilizing Multifunctional Spin-Hall MTJ Neurons

Abstract

We propose a new network architecture for standard spin-Hall magnetic tunnel junction-based spintronic neurons that allows them to compute multiple critical convolutional neural network functionalities simultaneously and in parallel, saving space and time. An approximation to the Rectified Linear Unit transfer function and the local pooling function are computed simultaneously with the convolution operation itself. A proof-of-concept simulation is performed on the MNIST dataset, achieving up to 98% accuracy at a cost of less than 1 nJ for all convolution, activation and pooling operations combined. The simulations are remarkably robust to thermal noise, performing well even with very small magnetic layers.

References in corpus (5)

• Spin torque switching with the giant spin Hall effect of tantalum
• Spin Transfer Torques
• Switching Distributions for Perpendicular Spin-Torque Devices within the Macrospin Approximation
• Non-volatile Spin Switch for Boolean and Non-Boolean Logic
• A Proposal for Energy-Efficient Cellular Neural Network based on Spintronic Devices