aiqbench analysis
This paper introduces counterfactual transport flows, a method for offline reinforcement learning that refines candidate trajectories using local preference pairs derived from historical data without extrapolating beyond observed distributions. By incorporating a refinement strength parameter, the approach balances behavioral preservation with performance gains on standard benchmarks like D4RL. The technique offers interpretable, instance-specific improvement paths, addressing a core challenge in leveraging logged data for policy optimization.
Counterfactual Transport Flows for Offline Conservative Trajectory Refinement
arXiv:2606.09115v1 Announce Type: new Abstract: Offline reinforcement learning (RL) offers a path to policy improvement from logged data alone, using historical returns or other measurable outcomes as world feedback. A key difficulty is improving observed behavior without extrapolating beyond what the offline data supports. We propose \emph{counterfactual transport flows}, a source-conditioned trajectory refinement framework for offline decision-making guided by world feedback. Given a low-feedback candidate trajectory, we construct local preference pairs from offline data by retrieving nearby trajectories in latent trajectory space with higher task-specific feedback, and use them as weak supervision for conservative refinement. The framework learns instance-specific refinement directions: at inference time, a refinement strength parameter controls how far the candidate trajectory is transported, enabling a trade-off between preserving the original behavior and applying stronger improvement. Experiments on D4RL benchmarks, including AntMaze and MuJoCo tasks, show that our method improves behavior from historical returns as world feedback, while providing interpretable trajectory-level refinement paths.
Abstract:Offline reinforcement learning (RL) offers a path to policy improvement from logged data alone, using historical returns or other measurable outcomes as world feedback. A key difficulty is improving observed behavior without extrapolating beyond what the offline data supports. We propose \emph{counterfactual transport flows}, a source-conditioned trajectory refinement framework for offline decision-making guided by world feedback. Given a low-feedback candidate trajectory, we construct local preference pairs from offline data by retrieving nearby trajectories in latent trajectory space with higher task-specific feedback, and use them as weak supervision for conservative refinement. The framework learns instance-specific refinement directions: at inference time, a refinement strength parameter controls how far the candidate trajectory is transported, enabling a trade-off between preserving the original behavior and applying stronger improvement. Experiments on D4RL benchmarks, including AntMaze and MuJoCo tasks, show that our method improves behavior from historical returns as world feedback, while providing interpretable trajectory-level refinement paths.
Machine Learning (cs.LG)
arXiv:2606.09115 [cs.LG]
(or arXiv:2606.09115v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2606.09115 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 244 words
Related coverage
Exact Fast Batch Simulation for Tabular Reinforcement Learning
arXiv · Oct 6, 2026
From Suboptimal to Optimal: One-Step Latent Flow Policies for Offline Reinforcement Learning
arXiv · Oct 15, 2026
Test-time Multi-agent Coordination by Decomposed Value Gradient Flow
arXiv · Oct 5, 2026
Reciprocity Separates Gradient Flow from Rotation in Conservative Physical Learning
arXiv · Sep 1, 2026