Very Fast, Approximate Counterfactual Explanations for Decision Forests
Miguel Á. Carreira-Perpinan, Suryabhan Singh Hada
Abstract
We consider finding a counterfactual explanation for a classification or regression forest, such as a random forest. This requires solving an optimization problem to find the closest input instance to a given instance for which the forest outputs a desired value. Finding an exact solution has a cost that is exponential on the number of leaves in the forest. We propose a simple but very effective approach: we constrain the optimization to input space regions populated by actual data points. The problem reduces to a form of nearest-neighbor search using a certain distance on a certain dataset. This has two advantages: first, the solution can be found very quickly, scaling to large forests and high-dimensional data, and enabling interactive use. Second, the solution found is more likely to be realistic in that it is guided towards high-density areas of input space.
BibTeX
@article{Carreira-Perpinan_Hada_2023, title={Very Fast, Approximate Counterfactual Explanations for Decision Forests}, volume={37}, url={https://ojs.aaai.org/index.php/AAAI/article/view/25848}, DOI={10.1609/aaai.v37i6.25848}, abstractNote={We consider finding a counterfactual explanation for a classification or regression forest, such as a random forest. This requires solving an optimization problem to find the closest input instance to a given instance for which the forest outputs a desired value. Finding an exact solution has a cost that is exponential on the number of leaves in the forest. We propose a simple but very effective approach: we constrain the optimization to input space regions populated by actual data points. The problem reduces to a form of nearest-neighbor search using a certain distance on a certain dataset. This has two advantages: first, the solution can be found very quickly, scaling to large forests and high-dimensional data, and enabling interactive use. Second, the solution found is more likely to be realistic in that it is guided towards high-density areas of input space.}, number={6}, journal={Proceedings of the AAAI Conference on Artificial Intelligence}, author={Carreira-Perpinan, Miguel Á. and Hada, Suryabhan Singh}, year={2023}, month={Jun.}, pages={6935-6943} }