SHE-MTJ Circuits for Convolutional Neural Networks

2020/07/16 by Andrew W. Stephan, Stephan, Andrew W., Steven J. Koester +1
Engineering · #Advanced Memory and Neural Computing #Emerging Technologies (cs.ET) #FOS: Computer and information sciences #FOS: Electrical engineering #Ferroelectric and Negative Capacitance Devices #Image and Video Processing (eess.IV) #Semiconductor materials and devices #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2007.08437

openalex publication_date 2020/07/16 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We report the performance characteristics of a notional Convolutional Neural Network based on the previously-proposed Multiply-Accumulate-Activate-Pool set, an MTJ-based spintronic circuit made to compute multiple neural functionalities in parallel. A study of image classification with the MNIST handwritten digits dataset using this network is provided via simulation. The effect of changing the weight representation precision, the severity of device process variation within the MAAP sets and the computational redundancy are provided. The emulated network achieves between 90 and 95% image classification accuracy at a cost of ~100 nJ per image.

Citations

• Gradient-based learning applied to document recognition

Related

• Convolutional Neural Networks Utilizing Multifunctional Spin-Hall MTJ Neurons
• Experimental Demonstration of Neuromorphic Network with STT MTJ Synapses
• Nonvolatile Spintronic Memory Cells for Neural Networks
• Power Optimizations in MTJ-based Neural Networks through Stochastic Computing
• Voltage Gated Domain Wall Magnetic Tunnel Junction-based Spiking Convolutional Neural Network