When to Trust Your Model: Model-Based Policy Optimization
This work addresses the problem of sample efficiency and scalability in reinforcement learning for researchers and practitioners, offering an incremental improvement over prior model-based methods.
The paper tackles the challenge of balancing data generation ease with model bias in model-based reinforcement learning by introducing a method that uses short model-generated rollouts from real data, achieving improved sample efficiency, matching asymptotic performance of model-free algorithms, and scaling to longer horizons.
Designing effective model-based reinforcement learning algorithms is difficult because the ease of data generation must be weighed against the bias of model-generated data. In this paper, we study the role of model usage in policy optimization both theoretically and empirically. We first formulate and analyze a model-based reinforcement learning algorithm with a guarantee of monotonic improvement at each step. In practice, this analysis is overly pessimistic and suggests that real off-policy data is always preferable to model-generated on-policy data, but we show that an empirical estimate of model generalization can be incorporated into such analysis to justify model usage. Motivated by this analysis, we then demonstrate that a simple procedure of using short model-generated rollouts branched from real data has the benefits of more complicated model-based algorithms without the usual pitfalls. In particular, this approach surpasses the sample efficiency of prior model-based methods, matches the asymptotic performance of the best model-free algorithms, and scales to horizons that cause other model-based methods to fail entirely.