A coupled oscillator field where the local learning rules (α_i, β_i, γ_i) themselves evolve as a second-order dynamical system, constrained by an identity curvature functional H[L], with Lyapunov-gated rollback of destructive meta-updates — unified in a single implementable architecture.
python python3 artificial-intelligence artificial-neural-networks artificial-intelligence-algorithms meta-learning active-inference non-equilibrium-dynamics oscillator-networks
-
Updated
Jun 28, 2026 - Jupyter Notebook