Self-Learning Machines Based on Hamiltonian Echo Backpropagation

2021/03/08 by Victor Lopez-Pastor, Víctor López-Pastor, Florian Marquardt · 2 voices · 2 citations
Computer Science · Physics and Astronomy · #Mechanical and Optical Resonators #Neural Networks and Reservoir Computing #cs.LG #nlin.AO #physics.data-an #physics.optics

paper · pdf · doi:10.1103/physrevx.13.031020

arxiv published 2021/03/08 · arxiv updated 2023/02/07 · openalex publication_date 2023/08/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/04

Abstract

A physical self-learning machine can be defined as a nonlinear dynamical system that can be trained on data (similar to artificial neural networks), but where the update of the internal degrees of freedom that serve as learnable parameters happens autonomously. In this way, neither external processing and feedback nor knowledge of (and control of) these internal degrees of freedom is required. We introduce a general scheme for self-learning in any time-reversible Hamiltonian system. We illustrate the training of such a self-learning machine numerically for the case of coupled nonlinear wave fields.

Citations

• Deep learning with coherent nanophotonic circuits
• Wave physics as an analog recurrent neural network
• Backpropagation and the brain
• Optimal bistable switching in nonlinear photonic crystals

Cited by

• Programmable metasurfaces for future photonic artificial intelligence
• Equilibrium Propagation for Learning in Lagrangian Dynamical Systems

Discussions

• Self-learning Machines based on Hamiltonian Echo Backpropagation (2021) [hn, 2 points, 0 comments]
• Self-learning Machines based on Hamiltonian Echo Backpropagation (2021) [hn, 1 points, 0 comments]