Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2402.10877 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917727403245568 |
|---|---|
| author | Richens, Jonathan Everitt, Tom |
| author_facet | Richens, Jonathan Everitt, Tom |
| contents | It has long been hypothesised that causal reasoning plays a fundamental role in robust and general intelligence. However, it is not known if agents must learn causal models in order to generalise to new domains, or if other inductive biases are sufficient. We answer this question, showing that any agent capable of satisfying a regret bound under a large set of distributional shifts must have learned an approximate causal model of the data generating process, which converges to the true causal model for optimal agents. We discuss the implications of this result for several research areas including transfer learning and causal inference. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_10877 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Robust agents learn causal world models Richens, Jonathan Everitt, Tom Artificial Intelligence Machine Learning It has long been hypothesised that causal reasoning plays a fundamental role in robust and general intelligence. However, it is not known if agents must learn causal models in order to generalise to new domains, or if other inductive biases are sufficient. We answer this question, showing that any agent capable of satisfying a regret bound under a large set of distributional shifts must have learned an approximate causal model of the data generating process, which converges to the true causal model for optimal agents. We discuss the implications of this result for several research areas including transfer learning and causal inference. |
| title | Robust agents learn causal world models |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2402.10877 |