Training deep physical neural networks with local physical information bottleneck

2026-02-10

AI Summary

This work addresses the lack of a general and efficient training methodology for deep physical neural networks that can operate robustly and scalably on real hardware across arbitrary physical dynamics. The authors propose the Physics-Informed Bottleneck (PIB) framework, which, for the first time, localizes the information bottleneck principle to individual computational units. By enforcing matrix-based information constraints and leveraging local learning rules, PIB enables intrinsic training without requiring digital surrogate models or comparative measurements. The approach unifies supervised, unsupervised, and reinforcement learning paradigms and is compatible with diverse physical hardware platforms—including memristive and photonic systems—while exhibiting strong fault tolerance and support for geographically distributed parallel training. This significantly enhances the practicality and scalability of physical neural networks.

Technology Category

• Machine Learning: Hardware-aware ML
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Deep learning has revolutionized modern society but faces growing energy and latency constraints. Deep physical neural networks (PNNs) are interconnected computing systems that directly exploit analog dynamics for energy-efficient, ultrafast AI execution. Realizing this potential, however, requires universal training methods tailored to physical intricacies. Here, we present the Physical Information Bottleneck (PIB), a general and efficient framework that integrates information theory and local learning, enabling deep PNNs to learn under arbitrary physical dynamics. By allocating matrix-based information bottlenecks to each unit, we demonstrate supervised, unsupervised, and reinforcement learning across electronic memristive chips and optical computing platforms. PIB also adapts to severe hardware faults and allows for parallel training via geographically distributed resources. Bypassing auxiliary digital models and contrastive measurements, PIB recasts PNN training as an intrinsic, scalable information-theoretic process compatible with diverse physical substrates.

Problem

• deep physical neural networks
• training method
• physical dynamics
• hardware faults
• information bottleneck

Innovation

• Physical Neural Networks
• Information Bottleneck
• Local Learning
• Hardware-Aware Training
• Analog Computing