Self-Composing Policies for Scalable Continual Reinforcement Learning

This work introduces a growable and modular neural network architecture that naturally avoids catastrophic forgetting and interference in con- tinual reinforcement learning. The structure of each module allows the selective combination of previous policies along with its internal policy, acceleratin...

Descripción completa

Detalles Bibliográficos
Autores: Malagon, M., Ceberio, J., Lozano, J.A.
Tipo de recurso: artículo
Estado:Versión publicada
Fecha de publicación:2025
País:España
Institución:Basque Center for Applied Mathematics (BCAM)
Repositorio:BIRD. BCAM's Institutional Repository Data
OAI Identifier:oai:bird.bcamath.org:20.500.11824/2042
Acceso en línea:http://hdl.handle.net/20.500.11824/2042
https://arxiv.org/abs/2506.14811v1
https://doi.org/10.48550/arXiv.2506.14811
Access Level:acceso abierto
Palabra clave:Self-Composing Policies
Scalable Continual Reinforcement
Descripción
Sumario:This work introduces a growable and modular neural network architecture that naturally avoids catastrophic forgetting and interference in con- tinual reinforcement learning. The structure of each module allows the selective combination of previous policies along with its internal policy, accelerating the learning process on the current task. Unlike previous growing neural network approaches, we show that the number of parame- ters of the proposed approach grows linearly with respect to the number of tasks, and does not sac- rifice plasticity to scale. Experiments conducted in benchmark continuous control and visual prob- lems reveal that the proposed approach achieves greater knowledge transfer and performance than alternative methods