Scalable Multi-Task Inverse Reinforcement Learning
Abstract
By learning transferable rewards, inverse reinforcement learning (IRL) enables counterfactual evaluation of agents under modified environments. Such transfer places strict requirements on coverage since target environments affect agents' state occupancy. We propose a multi-task IRL method that pools data across multiple agents with different rewards in the same environment under a low-rank assumption. In addition to alleviating coverage requirements, so each task need not visit every state as long as others do, the method offers scalable evaluation of multiple tasks under new environments as computationally intensive planning scales with rank rather than the number of tasks. We provide finite sample guarantees on reward recovery and on policy learning in new environments. Experiments show our method is robust to limited coverage, recovers rewards on and off of each task's support, transfers to target environments at lower regret than baselines, with its computational advantage over per-task methods widening as tasks grow.
Related
- TOPPO: Rethinking PPO for Multi-Task Reinforcement Learning with Critic Balancing
- Survival Reinforcement Learning: Toward Scalable Self-Supervised RL
- Adaptive Task Offloading Under Dynamic Topology via Graph Attention-Enhanced Reinforcement Learning for Layered UAV-Enabled Digital Twin Channel Modeling
- A Multi‐Task Learning Approach for Identifying Prognostic Variables Across Multiple Historical Clinical Trials
- Robust Multi-Task Learning for Principal Component Analysis
BibTeX
@misc{tran2026scalable,
title = {{Scalable Multi-Task Inverse Reinforcement Learning}},
author = {Allen Tran and Jia Wan and Nathan Kallus and Aurélien Bibaut},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.00758},
eprint = {2610.00758},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.00758},
}
RIS
TY - UNPB
TI - Scalable Multi-Task Inverse Reinforcement Learning
AU - Allen Tran
AU - Jia Wan
AU - Nathan Kallus
AU - Aurélien Bibaut
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2610.00758
UR - https://arxiv.org/abs/2610.00758
AB - By learning transferable rewards, inverse reinforcement learning (IRL) enables counterfactual evaluation of agents under modified environments. Such transfer places strict requirements on coverage since target environments affect agents' state occupancy. We propose a multi-task IRL method that pools data across multiple agents with different rewards in the same environment under a low-rank assumption. In addition to alleviating coverage requirements, so each task need not visit every state as long as others do, the method offers scalable evaluation of multiple tasks under new environments as computationally intensive planning scales with rank rather than the number of tasks. We provide finite sample guarantees on reward recovery and on policy learning in new environments. Experiments show our method is robust to limited coverage, recovers rewards on and off of each task's support, transfers to target environments at lower regret than baselines, with its computational advantage over per-task methods widening as tasks grow.
ER -