Noise-based Local Learning using Stochastic Magnetic Tunnel Junctions

Abstract

Brain-inspired learning in physical hardware has enormous potential to learn fast at minimal energy expenditure. One of the characteristics of biological learning systems is their ability to learn in the presence of various noise sources. Inspired by this observation, we introduce a novel noise-based learning approach for physical systems implementing multi-layer neural networks. Simulation results show that our approach allows for effective learning whose performance approaches that of the conventional effective yet energy-costly backpropagation algorithm. Using a spintronics hardware implementation, we demonstrate experimentally that learning can be achieved in a small network composed of physical stochastic magnetic tunnel junctions. These results provide a path towards efficient learning in general physical systems which embraces rather than mitigates the noise inherent in physical devices.

20 pages, 5 figures, submitted to Physical Review Applied

References in corpus (10)

• Stochastic p-bits for Invertible Logic
• Thermal skyrmion diffusion applied in probabilistic computing
• Brownian reservoir computing realized using geometrically confined skyrmions
• Demonstration of nanosecond operation in stochastic magnetic tunnel junctions
• Hardware-aware $insitu$ Boltzmann machine learning using stochastic magnetic tunnel junctions
• Demonstration of Decentralized, Physics-Driven Learning
• Supervised learning in physical networks: From machine learning to learning machines
• Evaluating spintronics-compatible implementations of Ising machines
• Self-learning Machines based on Hamiltonian Echo Backpropagation
• Gradient-Free Training of Recurrent Neural Networks using Random Perturbations