Building Reservoir Computing Hardware Using Low Energy-Barrier Magnetics

Samiran Ganguly, Avik W. Ghosh, arXiv preprint arXiv:2007.02766 (2020).
Full text


Biologically inspired recurrent neural networks, such as reservoir computers are of interest in designing spatio-temporal data processors from a hardware point of view due to the simple learning scheme and deep connections to Kalman filters. In this work we discuss using in-depth simulation studies a way to construct hardware reservoir computers using an analog stochastic neuron cell built from a low energy-barrier magnet based magnetic tunnel junction and a few transistors. This allows us to implement a physical embodiment of the mathematical model of reservoir computers. Compact implementation of reservoir computers using such devices may enable building compact, energy-efficient signal processors for standalone or in-situ machine cognition in edge devices.