LATIDIA · Robótica

DRAP: Plasticidad adaptativa dinámica de riesgo para el aprendizaje continuo en IA física

arXiv: 2611.05029v1Tipo de anuncio: Cross Resumen: El aprendizaje continuo en IA física difiere del aprendizaje continuo convencional porque se pueden acumular pequeños errores de política por paso a lo largo de una trayectoria. Cada desviación puede m

La noticia

arXiv:2611.05029v1 Announce Type: cross Abstract: Continual learning in physical AI differs from conventional continual learning because small per-step policy errors can accumulate over a trajectory. Each deviation may move the agent outside the training distribution, increase subsequent prediction errors, and ultimately cause closed-loop task failure. This work introduces Dynamics-Risk Adaptive Plasticity (DRAP), a replay-free method that estimates the historical importance of adaptable model tensors from sample-level action gradients weighted by local trajectory-amplification risk. A ridge-regularized local linear model is fitted to expert state-action transitions, and the largest singular value of its state-transition matrix provides a policy-independent, Lyapunov-inspired estimate of one-step perturbation amplification. After each task, risk-weighted squared gradients are accumulated