Holistic Approach to Measure Sample-level Adversarial Vulnerability and its Utility in Building Trustworthy Systems

Accepted at Human-centered Intelligent Services Safety and Trustworthy (CVPR 2022 - Workshop)

Project Website: https://sites.google.com/view/sample-adv-trustworthy/home

Steps to Run

• Install dependencies using pip3 install requirements.txt
• Run scripts/csv_CIFAR_PGD_train.sh for computing DDB, Flipping Frequency and Trust Score results across various attacks and .
• For analysing above obtained results use _code/trust_score.ipynb notebook
• For various plots and visualization refer _code/all_plotter.ipynb

Citation

If you find this work useful, please consider citing our work:

@inproceedings{
nayak2022_holistic,
title={Holistic Approach to Measure Sample-level Adversarial Vulnerability and its
Utility in Building Trustworthy Systems},
author={Nayak, G. K., Rawal, R., Lal, R., Patil, H., and Chakraborty, A.},
booktitle={Conference on Computer Vision and Pattern Recognition (CVPR) Workshop on Human-centered Intelligent Services Safety and Trustworthy},
year={2022}
}

Authors

Gaurav Kumar Nayak*, Ruchit Rawal*, Rohit Lal*, Himanshu Patil, Anirban Chakraborty (*Equal Contribution)