MUCH
MUlti Counterfactual Halton sampling
Counterfactual eXplanations (CEX) are spreading rapidly in the literature on explainable AI due to their ability to be easily understood and highly adaptable to data. In this repository you will found an original method for the extraction of multiple CEX from numerical data based on Halton sampling and Support Vector Data Description.
Examples may help to better understand the motivation behind this study and the importance of counterfactual reasoning in multiclass problems. In the field of health, several diseases present with different stages of severity (e.g., cancer, chronic obstructive pulmonary disease...) that can worsen drastically in a short time if not properly treated. In this case, multiclass counterfactuals can be a crucial instrument to monitor the stage of disease progression in order to detect minimal changes in the patient’s condition and apply appropriate countermeasures before the disease progresses to the next stage. Another example may instead involve the study of the transition of a phenomenon that develops over several stages (e.g., A, B, C, D). A counterfactual analysis can be useful to check for differences between different transitions (e.g., direct paths skipping intermediate transitions or progressive sequential paths).
The concept paper behind this algorithm is currently under review at the IEEE Transactions on Artificial Intelligence.
Enjoy !
Requirements
For the evaluation and visualization of the counterfactual explanations we used the following github repository:
Moses (2023). spider_plot (https://github.com/NewGuy012/spider_plot/releases/tag/19.4), GitHub. Retrieved January 17, 2023.
The datasets can be directly retrieved from Kaggle_
-) FIFA 2023: https://www.kaggle.com/datasets/cashncarry/fifa-23-complete-player-dataset
-) Stellar Classification: https://www.kaggle.com/fedesoriano/stellar-classification-dataset-sdss17
Citation pour cette source
Alberto Carlevaro (2024). MUCH (https://github.com/AlbiCarle/MUCH/releases/tag/v1.0.0), GitHub. Récupéré le .
Compatibilité avec les versions de MATLAB
Plateformes compatibles
Windows macOS LinuxTags
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!Découvrir Live Editor
Créez des scripts avec du code, des résultats et du texte formaté dans un même document exécutable.
Counterfactuals_Utilities
MC-SVDD
MultiFIFA
MultiStars
MultiFIFA
MultiStars
Version | Publié le | Notes de version | |
---|---|---|---|
1.0.0 |