A Method for Evaluating Hyperparameter Sensitivity in Reinforcement Learning
This addresses a scalability issue for RL practitioners by providing a tool to evaluate algorithm robustness, though it is incremental as it builds on existing sensitivity analysis concepts.
The paper tackles the problem of hyperparameter sensitivity in reinforcement learning by proposing a new empirical methodology to quantify how algorithm performance depends on hyperparameter tuning, and demonstrates that some performance improvements in PPO variants may stem from increased reliance on tuning.
The performance of modern reinforcement learning algorithms critically relies on tuning ever-increasing numbers of hyperparameters. Often, small changes in a hyperparameter can lead to drastic changes in performance, and different environments require very different hyperparameter settings to achieve state-of-the-art performance reported in the literature. We currently lack a scalable and widely accepted approach to characterizing these complex interactions. This work proposes a new empirical methodology for studying, comparing, and quantifying the sensitivity of an algorithm's performance to hyperparameter tuning for a given set of environments. We then demonstrate the utility of this methodology by assessing the hyperparameter sensitivity of several commonly used normalization variants of PPO. The results suggest that several algorithmic performance improvements may, in fact, be a result of an increased reliance on hyperparameter tuning.