Learning to reason about contextual knowledge for planning under uncertainty

Cheng Cui, Saeid Amiri, Yan Ding, Xingyue Zhan, Shiqi Zhang
Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence, PMLR 216:465-475, 2023.

Abstract

Sequential decision-making (SDM) methods enable AI agents to compute an action policy toward achieving long-term goals under uncertainty. Existing research has shown that contextual knowledge in declarative forms can be used for improving the performance of SDM methods. However, the contextual knowledge from people tends to be incomplete and sometimes inaccurate, which greatly limits the applicability of knowledge-based SDM methods. In this paper, we develop a novel algorithm for knowledge-based SDM, called PERIL, that learns from interaction experience to reason about contextual knowledge, as applied to urban driving scenarios. Experiments have been conducted using CARLA, a widely used autonomous driving simulator. Results demonstrate PERIL’s superiority in comparison to existing knowledge-based SDM baselines.

Cite this Paper

BibTeX
@InProceedings{pmlr-v216-cui23a, title = {Learning to reason about contextual knowledge for planning under uncertainty}, author = {Cui, Cheng and Amiri, Saeid and Ding, Yan and Zhan, Xingyue and Zhang, Shiqi}, booktitle = {Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence}, pages = {465--475}, year = {2023}, editor = {Evans, Robin J. and Shpitser, Ilya}, volume = {216}, series = {Proceedings of Machine Learning Research}, month = {31 Jul--04 Aug}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v216/cui23a/cui23a.pdf}, url = {https://proceedings.mlr.press/v216/cui23a.html}, abstract = {Sequential decision-making (SDM) methods enable AI agents to compute an action policy toward achieving long-term goals under uncertainty. Existing research has shown that contextual knowledge in declarative forms can be used for improving the performance of SDM methods. However, the contextual knowledge from people tends to be incomplete and sometimes inaccurate, which greatly limits the applicability of knowledge-based SDM methods. In this paper, we develop a novel algorithm for knowledge-based SDM, called PERIL, that learns from interaction experience to reason about contextual knowledge, as applied to urban driving scenarios. Experiments have been conducted using CARLA, a widely used autonomous driving simulator. Results demonstrate PERIL’s superiority in comparison to existing knowledge-based SDM baselines.} }
Endnote
%0 Conference Paper %T Learning to reason about contextual knowledge for planning under uncertainty %A Cheng Cui %A Saeid Amiri %A Yan Ding %A Xingyue Zhan %A Shiqi Zhang %B Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2023 %E Robin J. Evans %E Ilya Shpitser %F pmlr-v216-cui23a %I PMLR %P 465--475 %U https://proceedings.mlr.press/v216/cui23a.html %V 216 %X Sequential decision-making (SDM) methods enable AI agents to compute an action policy toward achieving long-term goals under uncertainty. Existing research has shown that contextual knowledge in declarative forms can be used for improving the performance of SDM methods. However, the contextual knowledge from people tends to be incomplete and sometimes inaccurate, which greatly limits the applicability of knowledge-based SDM methods. In this paper, we develop a novel algorithm for knowledge-based SDM, called PERIL, that learns from interaction experience to reason about contextual knowledge, as applied to urban driving scenarios. Experiments have been conducted using CARLA, a widely used autonomous driving simulator. Results demonstrate PERIL’s superiority in comparison to existing knowledge-based SDM baselines.
APA
Cui, C., Amiri, S., Ding, Y., Zhan, X. & Zhang, S.. (2023). Learning to reason about contextual knowledge for planning under uncertainty. Proceedings of the Thirty-Ninth Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 216:465-475 Available from https://proceedings.mlr.press/v216/cui23a.html.

Related Material