Coupled VO2 Oscillators Circuit as Analog First Layer Filter in Convolutional Neural Networks

Elisabetta Corti; Joaquin Antonio Cornejo Jimenez; Kham M Niang; John Robertson; Kirsten E Moselund; Bernd Gotsmann; Adrian M Ionescu; Siegfried Karg

doi:10.3389/fnins.2021.628254

Coupled VO₂ Oscillators Circuit as Analog First Layer Filter in Convolutional Neural Networks

Front Neurosci. 2021 Feb 11:15:628254. doi: 10.3389/fnins.2021.628254. eCollection 2021.

Authors

Elisabetta Corti¹, Joaquin Antonio Cornejo Jimenez¹, Kham M Niang², John Robertson², Kirsten E Moselund¹, Bernd Gotsmann¹, Adrian M Ionescu³, Siegfried Karg¹

Affiliations

¹ IBM Research Zürich, Rüschlikon, Switzerland.
² Department of Engineering, University of Cambridge, Cambridge, United Kingdom.
³ Nanoelectronic Devices Laboratory, École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland.

Abstract

In this work we present an in-memory computing platform based on coupled VO₂ oscillators fabricated in a crossbar configuration on silicon. Compared to existing platforms, the crossbar configuration promises significant improvements in terms of area density and oscillation frequency. Further, the crossbar devices exhibit low variability and extended reliability, hence, enabling experiments on 4-coupled oscillator. We demonstrate the neuromorphic computing capabilities using the phase relation of the oscillators. As an application, we propose to replace digital filtering operation in a convolutional neural network with oscillating circuits. The concept is tested with a VGG13 architecture on the MNIST dataset, achieving performances of 95% in the recognition task.

Keywords: convolutional neural networks; coupled oscillators; oscillatory neural network; pattern recognition; phase-encoding; relaxation oscillators; vanadium dioxide.