Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2402.01607 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866910678628958208 |
|---|---|
| author | Hao, Guang-Yuan Zhang, Jiji Huang, Biwei Wang, Hao Zhang, Kun |
| author_facet | Hao, Guang-Yuan Zhang, Jiji Huang, Biwei Wang, Hao Zhang, Kun |
| contents | Counterfactual reasoning is pivotal in human cognition and especially important for providing explanations and making decisions. While Judea Pearl's influential approach is theoretically elegant, its generation of a counterfactual scenario often requires too much deviation from the observed scenarios to be feasible, as we show using simple examples. To mitigate this difficulty, we propose a framework of \emph{natural counterfactuals} and a method for generating counterfactuals that are more feasible with respect to the actual data distribution. Our methodology incorporates a certain amount of backtracking when needed, allowing changes in causally preceding variables to minimize deviations from realistic scenarios. Specifically, we introduce a novel optimization framework that permits but also controls the extent of backtracking with a naturalness criterion. Empirical experiments demonstrate the effectiveness of our method. The code is available at https://github.com/GuangyuanHao/natural_counterfactuals. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2402_01607 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Natural Counterfactuals With Necessary Backtracking Hao, Guang-Yuan Zhang, Jiji Huang, Biwei Wang, Hao Zhang, Kun Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning Neural and Evolutionary Computing Methodology Counterfactual reasoning is pivotal in human cognition and especially important for providing explanations and making decisions. While Judea Pearl's influential approach is theoretically elegant, its generation of a counterfactual scenario often requires too much deviation from the observed scenarios to be feasible, as we show using simple examples. To mitigate this difficulty, we propose a framework of \emph{natural counterfactuals} and a method for generating counterfactuals that are more feasible with respect to the actual data distribution. Our methodology incorporates a certain amount of backtracking when needed, allowing changes in causally preceding variables to minimize deviations from realistic scenarios. Specifically, we introduce a novel optimization framework that permits but also controls the extent of backtracking with a naturalness criterion. Empirical experiments demonstrate the effectiveness of our method. The code is available at https://github.com/GuangyuanHao/natural_counterfactuals. |
| title | Natural Counterfactuals With Necessary Backtracking |
| topic | Artificial Intelligence Computer Vision and Pattern Recognition Machine Learning Neural and Evolutionary Computing Methodology |
| url | https://arxiv.org/abs/2402.01607 |