TrojanZoo provides a universal pytorch platform to conduct security researches (especially backdoor attacks/defenses) of image classification in deep learning.

Code for Data Poisoning Attacks Against Federated Learning Systems

Python package to create adversarial agents for membership inference attacks againts machine learning models

LISA Traffic Signs Dataset for Pytorch. For Classification. 32x32 images. I use this to reproduce the Activation Clustering Results.

Not All Poisons are Created Equal: Robust Training against Data Poisoning (ICML 2022)

Defenses for data poisoning attacks on machine learning-based NIDS

computing tf-idf ponderation with Hadoop MapReduce