RobustML

This repository contains the implementation source code of the following paper:

Analyzing and Improving the Robustness of Tabular Classifiers using Counterfactual Explanations

BibTeX:

@inproceedings{rasouli2021robustness,
               title={Analyzing and Improving the Robustness of Tabular Classifiers using Counterfactual Explanations},
               author={Rasouli, Peyman and Yu, Ingrid Chieh},
               booktitle={2021 20th IEEE International Conference on Machine Learning and Applications (ICMLA)},
               pages={1286-1293},
               year={2021},
               organization={IEEE},
               doi={10.1109/ICMLA52953.2021.00209}
}

Setup

1- Clone the repository using HTTP/SSH:

git clone https://github.com/peymanrasouli/RobustML

2- Install the following package containing GCC/g++ compilers and libraries:

sudo apt-get install build-essential

3- Create a conda virtual environment:

conda create -n RobustML python=3.8

4- Activate the conda environment:

conda activate RobustML

5- Standing in RobustML directory, install the requirements:

pip install -r requirements.txt

Reproducing the robustness analysis results

1- To reproduce the results of the robustness analysis of black-box models run:

python robustness_analysis_blackbox.py

2- To reproduce the results of the perturbation efficacy benchmark:

python perturbation_efficacy_benchmark.py

Reproducing the robustness improvement results

1- To reproduce the results of the robustness improvement of black-box models run:

python robustness_improvement_blackbox.py

2- To reproduce the results of the robustness improvement of neural network model run:

python robustness_improvement_nn.py

3- To reproduce the success rate results of baseline methods after robustness improvement run:

python robustness_improvement_success_rate.py