/road_evaluation

Source Code of the ROAD benchmark for feature attribution methods

Primary LanguageJupyter NotebookMIT LicenseMIT

road_evaluation

Source Code of the ROAD evaluation method for feature attribution methods.

Paper

Have a look at our paper:

Yao Rong, Tobias Leemann, Vadim Borisov, Gjergji Kasneci and Enkelejda Kasneci. "Evaluating Feature Attribution: An Information-Theoretic Perspective", 2022

Example

Check out the notebook RoadBenchmark.ipynb for an example.

Our benchmark can be included by using the files imputation.py and road.py.

We run the script using:

-Python=3.8.8

-Pytorch=1.8.1

-torchvision=0.9

Credits

Please cite us if you use our code or ressources in your own work.