S-CFE: Simple Counterfactual Explanations

This repository contains the code for the paper "S-CFE: Simple Counterfactual Explanations". To run the code, execute the main.py file with the desired arguments, e.g.

python main.py --dataset Wine --alg SCFE_KDE --stepsize 0.01 --sparsity 2 --plausibility 5 --n_neighbors 3 --prox zero --model DNN --resdir ./Results/

where sparsity is the parameter beta in the paper and plausibility is the parameter theta in the paper.

Requirements

PyTorch, TorchVision, scikit-learn, Pandas, Numpy