Off-policy evaluation in reinforcement learning offers the chance of using observational data to improve future outcomes in domains such as healthcare and education, but safe …
Offline-to-online Reinforcement Learning (O2O RL) aims to improve the performance of offline pretrained policy using only a few online samples. Built on offline RL algorithms, most …
D Xu, Y Ye, C Ruan, B Yang - Proceedings of the AAAI Conference on …, 2022 - ojs.aaai.org
Off-policy learning plays a pivotal role in optimizing and evaluating policies prior to the online deployment. However, during the real-time serving, we observe varieties of …
S Liu, S Zhang - Forty-first International Conference on Machine … - openreview.net
Most reinforcement learning practitioners evaluate their policies with online Monte Carlo estimators for either hyperparameter tuning or testing different algorithmic design choices …
M Hao, P Su, L Hu, Z Szabo, Q Zhao, C Shi - arXiv preprint arXiv …, 2024 - arxiv.org
Off-policy evaluation (OPE) is crucial for evaluating a target policy's impact offline before its deployment. However, achieving accurate OPE in large state spaces remains challenging …
L Da, P Jenkins, T Schwantes, J Dotson… - Proceedings of the AAAI …, 2024 - ojs.aaai.org
In practice, it is essential to compare and rank candidate policies offline before real-world deployment for safety and reliability. Prior work seeks to solve this offline policy ranking …
Y Luo, T Ji, F Sun, J Zhang, H Xu, X Zhan - arXiv preprint arXiv …, 2024 - arxiv.org
Off-policy reinforcement learning (RL) has achieved notable success in tackling many complex real-world tasks, by leveraging previously collected data for policy learning …
J Cao, Q Liu, F Zhu, Q Fu, S Zhong - Information Sciences, 2021 - Elsevier
The problem of off-policy evaluation (OPE) has long been advocated as one of the foremost challenges in reinforcement learning. Gradient-based and emphasis-based temporal …