Selective Dyna-style Planning Under Limited Model Capacity
This work addresses model-based reinforcement learning for agents needing to avoid harmful planning, but it is incremental as it builds on prior uncertainty estimation methods.
The paper tackles the problem of planning with imperfect models in reinforcement learning by proposing selective planning based on model inadequacy, showing that heteroscedastic regression can detect complementary predictive uncertainty compared to parameter uncertainty methods.
In model-based reinforcement learning, planning with an imperfect model of the environment has the potential to harm learning progress. But even when a model is imperfect, it may still contain information that is useful for planning. In this paper, we investigate the idea of using an imperfect model selectively. The agent should plan in parts of the state space where the model would be helpful but refrain from using the model where it would be harmful. An effective selective planning mechanism requires estimating predictive uncertainty, which arises out of aleatoric uncertainty, parameter uncertainty, and model inadequacy, among other sources. Prior work has focused on parameter uncertainty for selective planning. In this work, we emphasize the importance of model inadequacy. We show that heteroscedastic regression can signal predictive uncertainty arising from model inadequacy that is complementary to that which is detected by methods designed for parameter uncertainty, indicating that considering both parameter uncertainty and model inadequacy may be a more promising direction for effective selective planning than either in isolation.