AutoRL Hyperparameter Landscapes

TL;DR

This paper investigates the changing nature of hyperparameter landscapes in reinforcement learning, providing empirical evidence that supports dynamic hyperparameter adjustment during training across various algorithms and environments.

Contribution

It introduces a method to analyze hyperparameter landscapes over time in AutoRL, revealing their strong variation and supporting dynamic hyperparameter tuning.

Findings

Hyperparameter landscapes vary significantly over training time.

Dynamic adjustment of hyperparameters is justified by landscape variability.

Landscape analysis offers new insights for AutoRL optimization strategies.

Abstract

Although Reinforcement Learning (RL) has shown to be capable of producing impressive results, its use is limited by the impact of its hyperparameters on performance. This often makes it difficult to achieve good results in practice. Automated RL (AutoRL) addresses this difficulty, yet little is known about the dynamics of the hyperparameter landscapes that hyperparameter optimization (HPO) methods traverse in search of optimal configurations. In view of existing AutoRL approaches dynamically adjusting hyperparameter configurations, we propose an approach to build and analyze these hyperparameter landscapes not just for one point in time but at multiple points in time throughout training. Addressing an important open question on the legitimacy of such dynamic AutoRL approaches, we provide thorough empirical evidence that the hyperparameter landscapes strongly vary over time across representative algorithms from RL literature (DQN, PPO, and SAC) in different kinds of environments (Cartpole, Bipedal Walker, and Hopper) This supports the theory that hyperparameters should be dynamically adjusted during training and shows the potential for more insights on AutoRL problems that can be gained through landscape analyses. Our code can be found at https://github.com/automl/AutoRL-Landscape