Bayesian Reinforcement Learning with Behavioral Feedback
Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence. Pages 1571-1576.
Abstract
In the standard reinforcement learning setting, the agent learns optimal policy solely from state transitions and rewards from the environment. We consider an extended setting where a trainer additionally provides feedback on the actions executed by the agent. This requires appropriately incorporating the feedback, even when the feedback is not necessarily accurate. In this paper, we present a Bayesian approach to this extended reinforcement learning setting. Specifically, we extend Kalman Temporal Difference learning to compute the posterior distribution over Q-values given the state transitions and rewards from the environment as well as the feedback from the trainer. Through experiments on standard reinforcement learning tasks, we show that learning performance can be significantly improved even with inaccurate feedback.
@inproceedings{ijcai2016p225,
title = {Bayesian Reinforcement Learning with Behavioral Feedback},
author = {Hong, Teakgyu and Lee, Jongmin and Kim, Kee-Eung and Ortega, Pedro A. and Lee, Daniel D.},
booktitle = {Proceedings of the Twenty-Fifth International Joint Conference on
Artificial Intelligence, {IJCAI-16}},
publisher = {AAAI Press / International Joint Conferences on Artificial Intelligence},
editor = {Subbarao Kambhampati},
pages = {1571--1576},
year = {2016},
month = {7},
url = {https://ijcai.org/proceedings/2016/225},
}
TY - CPAPER TI - Bayesian Reinforcement Learning with Behavioral Feedback AU - Hong, Teakgyu AU - Lee, Jongmin AU - Kim, Kee-Eung AU - Ortega, Pedro A. AU - Lee, Daniel D. T2 - Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence, IJCAI-16 PB - AAAI Press / International Joint Conferences on Artificial Intelligence PY - 2016 SP - 1571 EP - 1576 UR - https://ijcai.org/proceedings/2016/225 ER -Copied
Machine-readable versions of this paper, for tools and AI agents:
