Conditional Energy-Based Models for Implicit Policies: The Gap between Theory and Practice

2022/07/12 by Duy-Nguyen Ta, Ta, Duy-Nguyen, Eric Cousineau +5 · 1 citation
Psychology · #Behavioral and Psychological Studies #FOS: Computer and information sciences #Machine Learning (cs.LG) #Robotics (cs.RO)

paper · pdf · doi:10.48550/arxiv.2207.05824

openalex publication_date 2022/07/12 · openalex created_date 2022/07/15 · openalex updated_date 2026/07/28

Abstract

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.

Cited by

• Time-Unified Diffusion Policy with Action Discrimination for Robotic Manipulation

Related

• Implicit Behavioral Cloning
• When to Trust Your Model: Model-Based Policy Optimization
• Conditional Importance Sampling for Off-Policy Learning
• On Feature Diversity in Energy-based Models
• Implicit Policy for Reinforcement Learning