Event
Research analyzes PPO policy standard deviation and loss of plasticity
1 reports · 1 sources
Andrea Vasco Grieco and other authors published a paper analyzing loss of plasticity (LoP) in isotropic Gaussian policies in proximal policy optimization (PPO). The conclusion is that the policy standard deviation σ significantly affects the degree of LoP, and theoretical and experimental support is provided. The experiments include LoP benchmark comparisons and real policy deployment on a 7-degree-of-freedom robotic arm; based on this, the authors explain why common LoP mitigation methods are effective.
Report timeline
- Study Reveals Effect of Policy Standard Deviation on Loss of Plasticity in Online Reinforcement Learning
The paper analyzes loss of plasticity (LoP) in isotropic Gaussian policies within Proximal Policy Optimization (PPO), noting that the policy standard deviation σ significantly affects the extent of LoP, with theoretical and experimental support. The experiments include LoP benchmark comparisons and real-world policy deployment on a 7-DoF robotic arm; the authors also use these findings to explain why common LoP mitigation methods are effective.
arXiv cs.RO ·