Conditional Energy-Based Models for Implicit Policies: The Gap between Theory and Practice
This work addresses methodological pitfalls for researchers applying EBMs in reinforcement learning, but it is incremental as it clarifies existing theory rather than introducing new techniques.
The paper investigates the gap between theory and practice in using conditional energy-based models (EBM) for behavior-cloned policies, highlighting issues with generalization and emphasizing the Maximum Mutual Information principle as crucial for regression tasks.
We present our findings in the gap between theory and practice of using conditional energy-based models (EBM) as an implicit representation for behavior-cloned policies. We also clarify several subtle, and potentially confusing, details in previous work in an attempt to help future research in this area. We point out key differences between unconditional and conditional EBMs, and warn that blindly applying training methods for one to the other could lead to undesirable results that do not generalize well. Finally, we emphasize the importance of the Maximum Mutual Information principle as a necessary condition to achieve good generalization in conditional EBMs as implicit models for regression tasks.