3 papers
cs.LG2026
Beyond Single-Model Optimization: Preserving Plasticity in Continual Reinforcement Learning
Lute Lillo, Nick Cheney
Continual reinforcement learning must balance retention with adaptation, yet many methods still rely on \emph{single-model preservation}, committing to one evolving policy as the m…
cs.LG2025
How Weight Resampling and Optimizers Shape the Dynamics of Continual Learning and Forgetting in Neural Networks
Lapo Frati, Neil Traft, Jeff Clune +1
Recent work in continual learning has highlighted the beneficial effect of resampling weights in the last layer of a neural network (``zapping"). Although empirical results demonst…
cs.LG2024
Reset It and Forget It: Relearning Last-Layer Weights Improves Continual and Transfer Learning
Lapo Frati, Neil Traft, Jeff Clune +1
This work identifies a simple pre-training mechanism that leads to representations exhibiting better continual and transfer learning. This mechanism -- the repeated resetting of we…