Harry24k/PGD-pytorch

A pytorch implementation of "Towards Deep Learning Models Resistant to Adversarial Attacks"

47
/ 100
Emerging

This project helps machine learning engineers and researchers evaluate how vulnerable their image classification models are to malicious inputs. It takes a pre-trained image classification model and an image, then generates a slightly modified (adversarial) image that tricks the model into misclassifying it. This is useful for understanding and improving the robustness of AI systems in security-sensitive applications.

159 stars. No commits in the last 6 months.

Use this if you need to test the resilience of your deep learning image classification models against subtle, intentionally designed attacks that could cause misclassifications.

Not ideal if you are looking for a general-purpose library of various adversarial attacks, as this project specifically implements the PGD attack and is no longer updated.

deep-learning-security model-robustness adversarial-machine-learning image-recognition-defense AI-safety
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 10 / 25
Maturity 16 / 25
Community 21 / 25

How are scores calculated?

Stars

159

Forks

40

Language

Jupyter Notebook

License

MIT

Last pushed

Sep 04, 2019

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/Harry24k/PGD-pytorch"

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