arXiv cs.AI
· Papers
NeuMoSync: End-to-End Neuromodulatory Control for Plasticity and Adaptability in Continual Learning
arXiv:2608.04358v1 Announce Type: new Abstract: Continual learning (CL) requires models to learn tasks sequentially, yet deep neural networks often suffer from plasticity loss and poor knowledge transfer, which can impede their long-term adaptability. Drawing high-level inspiration from global neuromodulatory mechanism