On The Sample Efficiency Of Inverse Dynamics Models For Semi-supervised Imitation Learning

Abstract

Semi-supervised imitation learning (SSIL) consists in learning a policy from a small dataset of action-labeled trajectories and a much larger dataset of action-free trajectories. Some SSIL methods learn an inverse dynamics model (IDM) to predict the action from the current state and the next state. An IDM can act as a policy when paired with a video model (VM-IDM) or as a label generator to perform behavior cloning on action-free data (IDM labeling). In this work, we first show that VM-IDM and IDM labeling learn the same policy in a limit case, which we call the IDM-based policy. We then argue that the previously observed advantage of IDM-based policies over behavior cloning is due to the superior sample efficiency of IDM learning, which we attribute to two causes: (i) the ground-truth IDM tends to be contained in a lower complexity hypothesis class relative to the expert policy, and (ii) the ground-truth IDM is often less stochastic than the expert policy. We argue these claims based

Stats

• citations0
• S2 citations—
• github stars0
• HF likes0
• heat score0.00
• arxiv keymorin2026on

Related papers

• Deterministic And Discriminative Imitation (d2-imitation): Revisiting Adversarial Imitation For Sample Efficiency (2021)0.00
• Imitation Learning From Observations By Minimizing Inverse Dynamics Disagreement (2019)0.00
• State-only Imitation With Transition Dynamics Mismatch (2020)0.00
• Iq-learn: Inverse Soft-q Learning For Imitation (2021)0.00
• Mimicking Better By Matching The Approximate Action Distribution (2023)0.00
• Good Better Best: Self-motivated Imitation Learning For Noisy Demonstrations (2023)0.00
• Inverse Reinforcement Learning With Simultaneous Estimation Of Rewards And Dynamics (2016)0.00
• Efficient Inference For Inverse Reinforcement Learning And Dynamic Discrete Choice Models (2025)0.00