cmu-sei/feud
AI Division, Reverse Engineering CNN Trojans
This project helps security researchers and AI assurance specialists understand and reverse-engineer poisoned CNN models. You input a compromised Convolutional Neural Network and a set of salient images for the target class, and it outputs a refined, human-interpretable description and image of the hidden 'trojan' trigger. This helps you identify and mitigate malicious manipulations within AI systems.
No commits in the last 6 months.
Use this if you need to investigate and characterize a 'trojan' or adversarial patch embedded within a Convolutional Neural Network.
Not ideal if you are looking for a general-purpose CNN interpretability tool for benign models or a solution to remove trojans automatically.
Stars
9
Forks
1
Language
Python
License
—
Last pushed
Apr 09, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/cmu-sei/feud"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
obss/sahi
Framework agnostic sliced/tiled inference + interactive ui + error analysis plots
tensorflow/tcav
Code for the TCAV ML interpretability project
MAIF/shapash
🔅 Shapash: User-friendly Explainability and Interpretability to Develop Reliable and Transparent...
TeamHG-Memex/eli5
A library for debugging/inspecting machine learning classifiers and explaining their predictions
csinva/imodels
Interpretable ML package 🔍 for concise, transparent, and accurate predictive modeling...