serre-lab/Adversarial-Alignment
Scaling-up deep neural networks to improve their performance on ImageNet makes them more tolerant to adversarial attacks, but successful attacks on these models are misaligned with human perception.
This project helps researchers and practitioners in computer vision understand how robust deep neural networks are to adversarial attacks, specifically focusing on whether these attacks are perceptible to humans. It takes in various deep neural network models and an image dataset with human attention data to evaluate the strength and human-perceptibility of adversarial attacks. The primary users are AI researchers, computer vision engineers, and those working on secure or interpretable AI systems.
No commits in the last 6 months.
Use this if you are evaluating the vulnerability of your image recognition models to adversarial attacks and need to understand if those attacks are visually noticeable to humans.
Not ideal if you are looking for a tool to generate robust models directly, rather than analyze the robustness and human alignment of existing models.
Stars
7
Forks
1
Language
Jupyter Notebook
License
MIT
Category
Last pushed
Jun 28, 2023
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/serre-lab/Adversarial-Alignment"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
Trusted-AI/adversarial-robustness-toolbox
Adversarial Robustness Toolbox (ART) - Python Library for Machine Learning Security - Evasion,...
bethgelab/foolbox
A Python toolbox to create adversarial examples that fool neural networks in PyTorch, TensorFlow, and JAX
DSE-MSU/DeepRobust
A pytorch adversarial library for attack and defense methods on images and graphs
cleverhans-lab/cleverhans
An adversarial example library for constructing attacks, building defenses, and benchmarking both
BorealisAI/advertorch
A Toolbox for Adversarial Robustness Research