Self-learning Machines based on Hamiltonian Echo Backpropagation
This addresses the challenge of autonomous learning in physical systems, potentially enabling more efficient and integrated machine learning hardware, though it appears incremental as it builds on existing Hamiltonian concepts.
The paper tackles the problem of training physical self-learning machines without external feedback or control, introducing a scheme for self-learning in time-reversible Hamiltonian systems and demonstrating it numerically for coupled nonlinear wave fields.
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.