Physical Neural Networks with Self-Learning Capabilities

TL;DR

This paper reviews recent advances in physical neural networks that utilize intrinsic physical processes for self-learning, aiming to create hardware with embedded adaptive capabilities beyond traditional digital systems.

Contribution

It provides a comprehensive overview of physical self-learning implementations, discussing strategies, challenges, and progress in developing intelligent hardware with self-organizing dynamics.

Findings

Progress in physical self-learning systems across various physical platforms

Discussion of prevailing learning strategies for physical neural networks

Highlighting challenges and future directions in understanding physical learning mechanisms

Abstract

Physical neural networks are artificial neural networks that mimic synapses and neurons using physical systems or materials. These networks harness the distinctive characteristics of physical systems to carry out computations effectively, potentially surpassing the constraints of conventional digital neural networks. A recent advancement known as ``physical self-learning'' aims to achieve learning through intrinsic physical processes rather than relying on external computations. This article offers a comprehensive review of the progress made in implementing physical self-learning across various physical systems. Prevailing learning strategies are discussed that contribute to the realization of physical self-learning. Despite challenges in understanding fundamental mechanism of learning, this work highlights the progress towards constructing intelligent hardware from the ground up, incorporating embedded self-organizing and self-adaptive dynamics in physical systems.