Experimental demonstration of a robust training method for strongly defective neuromorphic hardware

Published

December 11, 2023

Author(s)

William Borders, Advait Madhavan, Matthew Daniels, Vasileia Georgiou, Martin Lueker-Boden, Tiffany Santos, Patrick Braganca, Mark Stiles, Jabez J. McClelland, Brian Hoskins

Abstract

Neural networks are increasing in scale and sophistication, catalyzing the need for efficient hardware. An inevitability when transferring neural networks to hardware is that non-idealities impact performance. Hardware-aware training, where non-idealities are accounted for during training is one way to recover performance, but at the cost of generality. In this work, we demonstrate a binary neural network consisting of an array of 20,000 magnetic tunnel junctions (MTJ) integrated on complementary metal-oxide-semiconductor (CMOS) chips. With 36 dies, we show that even a few defects can degrade the performance of neural networks. We demonstrate hardware-aware training and show that performance recovers close to ideal networks. We then introduce a robust method – statistics-aware training – that compensates for defects regardless of their specific configuration. When evaluated on the MNIST dataset, statistics-aware solutions differ from software-baselines by only 2 %.

Proceedings Title

Will not elect to publish in proceedings
The 35th Magnetic Recording Conference

Conference Dates

August 5-7, 2024

Conference Location

Berkeley, CA, US

Pub Type

Conferences

Keywords

Neuromorphic computing, magnetic tunnel junctions, neural network training, in-memory computing

Citation

Borders, W. , Madhavan, A. , Daniels, M. , Georgiou, V. , Lueker-Boden, M. , Santos, T. , Braganca, P. , Stiles, M. , McClelland, J. and Hoskins, B. (2023), Experimental demonstration of a robust training method for strongly defective neuromorphic hardware, Will not elect to publish in proceedings The 35th Magnetic Recording Conference, Berkeley, CA, US (Accessed July 17, 2024)