liuxuannan/Stochastic-Gradient-Aggregation

Official implementation of the ICCV2023 paper: Enhancing Generalization of Universal Adversarial Perturbation through Gradient Aggregation

24
/ 100
Experimental

This project helps machine learning engineers and researchers create Universal Adversarial Perturbations (UAPs) that are highly effective across many images and different AI models. It takes an existing image classification dataset and a trained AI model, then generates a small, imperceptible distortion (UAP). This UAP can reliably trick a target AI model into misclassifying most images it processes, making the model vulnerable to attack.

No commits in the last 6 months.

Use this if you need to generate highly generalizable Universal Adversarial Perturbations (UAPs) to test the robustness and security of your image classification AI models against sophisticated attacks.

Not ideal if you are looking to create instance-specific adversarial examples, or if your primary goal is to improve the base accuracy of your AI model rather than evaluating its vulnerability.

AI-security adversarial-machine-learning computer-vision model-robustness image-classification-auditing
No License Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 7 / 25
Maturity 8 / 25
Community 9 / 25

How are scores calculated?

Stars

28

Forks

3

Language

Python

License

Last pushed

Aug 17, 2023

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/liuxuannan/Stochastic-Gradient-Aggregation"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.