The official implementation of the CCS'23 paper, Narcissus clean-label backdoor attack -- only takes THREE images to poison a face recognition dataset in a clean-label way and achieves a 99.89% attack success rate.

Tensorflow implementation of TrialAttack (Triple Adversarial Learning for Influence based Poisoning Attack in Recommender Systems. KDD 2021)

Tensorflow implementation of APT (Fight Fire with Fire: Towards Robust Recommender Systems via Adversarial Poisoning Training. SIGIR 2021)

FedAnil is a secure blockchain-enabled Federated Deep Learning Model to address non-IID data and privacy concerns. This repo hosts a simulation for FedAnil written in Python.

FedAnil+ is a novel lightweight, and secure Federated Deep Learning Model to address non-IID data, privacy concerns, and communication overhead. This repo hosts a simulation for FedAnil+ written in Python.

FedAnil++ is a Privacy-Preserving and Communication-Efficient Federated Deep Learning Model to address non-IID data, privacy concerns, and communication overhead. This repo hosts a simulation for FedAnil++ written in Python.

Poisoning attack methods against adversarial training algorithms

Test tool to simulate two types of poisoning attack on AI model

Indirect Invisible Poisoning Attacks on Domain Adaptation

Test tool to simulate defense from poisoning attack on AI model

[Usenix Security 2025] On the Proactive Generation of Unsafe Images From Text-To-Image Models Using Benign Prompts

Tensorflow implementation of TrialAttack (Triple Adversarial Learning for Influence based Poisoning Attack in Recommender Systems. KDD 2021)

This project uses Python and machine learning to classify plant species as poisonous or non-poisonous. It aims to provide an efficient way to identify safe and harmful plants, useful for botanists, hikers, and the agricultural sector.

Tensorflow implementation of TrialAttack (Triple Adversarial Learning for Influence based Poisoning Attack in Recommender Systems. KDD 2021)