Parseval Regularization for Continual Reinforcement Learning
This addresses trainability issues for RL agents in sequential task settings, though it is incremental as it adapts an existing regularization technique to a new context.
The paper tackled the problem of loss of plasticity and trainability in continual reinforcement learning by applying Parseval regularization to maintain weight orthogonality, resulting in significant improvements on gridworld, CARL, and MetaWorld tasks.
Loss of plasticity, trainability loss, and primacy bias have been identified as issues arising when training deep neural networks on sequences of tasks -- all referring to the increased difficulty in training on new tasks. We propose to use Parseval regularization, which maintains orthogonality of weight matrices, to preserve useful optimization properties and improve training in a continual reinforcement learning setting. We show that it provides significant benefits to RL agents on a suite of gridworld, CARL and MetaWorld tasks. We conduct comprehensive ablations to identify the source of its benefits and investigate the effect of certain metrics associated to network trainability including weight matrix rank, weight norms and policy entropy.