Noise-Aware Training of Neuromorphic Dynamic Device Networks

TL;DR

This paper introduces a noise-aware training method for neuromorphic device networks using Neural-SDEs, enabling efficient, robust learning of complex dynamic tasks without detailed device models.

Contribution

It presents a novel Neural-SDE based approach for training physical device networks that accounts for noise and device dynamics, reducing data needs and improving robustness.

Findings

Effective on spintronic device networks

Improves temporal classification and regression tasks

Reduces training data requirements

Abstract

Physical computing has the potential to enable widespread embodied intelligence by leveraging the intrinsic dynamics of complex systems for efficient sensing, processing, and interaction. While individual devices provide basic data processing capabilities, networks of interconnected devices can perform more complex and varied tasks. However, designing networks to perform dynamic tasks is challenging without physical models and accurate quantification of device noise. We propose a novel, noise-aware methodology for training device networks using Neural Stochastic Differential Equations (Neural-SDEs) as differentiable digital twins, accurately capturing the dynamics and associated stochasticity of devices with intrinsic memory. Our approach employs backpropagation through time and cascade learning, allowing networks to effectively exploit the temporal properties of physical devices. We validate our method on diverse networks of spintronic devices across temporal classification and regression benchmarks. By decoupling the training of individual device models from network training, our method reduces the required training data and provides a robust framework for programming dynamical devices without relying on analytical descriptions of their dynamics.