ML Anthology

Abstraction for Deep Reinforcement Learning

Murray Shanahan, Melanie Mitchell
IJCAI 2022 pp. 5588-5596
doi:10.24963/IJCAI.2022/780 /ijcai/2022/shanahan2022ijcai-abstraction/

Abstract

We characterise the problem of abstraction in the context of deep reinforcement learning. Various well established approaches to analogical reasoning and associative memory might be brought to bear on this issue, but they present difficulties because of the need for end-to-end differentiability. We review developments in AI and machine learning that could facilitate their adoption.

PDF IJCAI Semantic Scholar

Cite

Text

Shanahan and Mitchell. "Abstraction for Deep Reinforcement Learning." International Joint Conference on Artificial Intelligence, 2022. doi:10.24963/IJCAI.2022/780

Markdown

[Shanahan and Mitchell. "Abstraction for Deep Reinforcement Learning." International Joint Conference on Artificial Intelligence, 2022.](https://mlanthology.org/ijcai/2022/shanahan2022ijcai-abstraction/) doi:10.24963/IJCAI.2022/780

BibTeX

@inproceedings{shanahan2022ijcai-abstraction,
  title     = {{Abstraction for Deep Reinforcement Learning}},
  author    = {Shanahan, Murray and Mitchell, Melanie},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year      = {2022},
  pages     = {5588-5596},
  doi       = {10.24963/IJCAI.2022/780},
  url       = {https://mlanthology.org/ijcai/2022/shanahan2022ijcai-abstraction/}
}