Deterministic and discriminative imitation (D2-imitation): revisiting adversarial imitation for sample efficiency

Conference item

Sample efficiency is crucial for imitation learning methods to be applicable in real-world applications. Many studies improve sample efficiency by extending adversarial imitation to be off-policy regardless of the fact that these off-policy extensions could either change the original objective or involve complicated optimization. We revisit the foundation of adversarial imitation and propose an off-policy sample efficient approach that requires no adversarial training or min-max optimization. Our formulation capitalizes on two key insights: (1) the similarity between the Bellman equation and the stationary state-action distribution equation allows us to derive a novel temporal difference (TD) learning approach; and (2) the use of a deterministic policy simplifies the TD learning. Combined, these insights yield a practical algorithm, Deterministic and Discriminative Imitation (D2-Imitation), which oper- ates by first partitioning samples into two replay buffers and then learning a deterministic policy via off-policy reinforcement learning. Our empirical results show that D2-Imitation is effective in achieving good sample efficiency, outperforming several off-policy extension approaches of adversarial imitation on many control tasks.

Authors: Sun, M; Devlin, S; Hofmann, K; Whiteson, S

• Publication status: Published
• Publisher: Association for the Advancement of Artificial Intelligence
• Journal: Proceedings of the 2022 AAAI Conference on Artificial Intelligence
• Volume: 36
• Issue: 8
• Pages: 8378-8385
• Publication date: 2022-06-28
• Acceptance date: 2022-01-01
• Event title: 2022 AAAI Conference on Artificial Intelligence
• DOI: 10.1609/aaai.v36i8.20813
• Language: English
• Keywords: FFR