Preprint
Aug 2026
Free-Energy-Gated Plasticity for Real-Time Online Motor Learning in Physical Human-Robot Interaction
The Predictive-Coding-inspired Variational Recurrent Neural Network is extended to continuously adapt its synaptic weights and Free-Energy-Gated Plasticity (FEGP) is proposed, which regulates the effective learning rate according to variational free energy.
Hiroki Sawada, Jun Tani
· 1 citation