pmzzs

Pinned Repositories

Narcissus
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.
Language:Python10010
Meta-Sift
The official implementation of USENIX Security'23 paper "Meta-Sift" -- Ten minutes or less to find a 1000-size or larger clean subset on poisoned dataset.
Language:Python154
JigMark
Language:Python50
meta_sift_artifacts
meta_sift_artifacts
Language:Python20
Meta-Sift
The official implementation of Meta-Sift -- Ten minutes or less to find a 1000-size or larger clean subset on any poisoned dataset.
Language:Python00
Narcissus-backdoor-attack
The official implementation of 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.
Language:Python00
standard-readme
A standard style for README files
Language:Shell00
WMD
WaterMark Detector (WMD) is a versatile and high-performance tool for detecting invisible watermarks in images without prior knowledge of the watermarking techniques used.
Language:Jupyter Notebook00
ASSET
This repository is the official implementation of the paper "ASSET: Robust Backdoor Data Detection Across a Multiplicity of Deep Learning Paradigms." ASSET achieves state-of-the-art reliability in detecting poisoned samples in end-to-end supervised learning/ self-supervised learning/ transfer learning.
Language:Python150

pmzzs's Repositories

pmzzs/WMD
WaterMark Detector (WMD) is a versatile and high-performance tool for detecting invisible watermarks in images without prior knowledge of the watermarking techniques used.
Language:Jupyter Notebook
pmzzs/JigMark
Language:Python5
pmzzs/meta_sift_artifacts
meta_sift_artifacts
Language:Python2
pmzzs/Meta-Sift
The official implementation of Meta-Sift -- Ten minutes or less to find a 1000-size or larger clean subset on any poisoned dataset.
pmzzs/Narcissus-backdoor-attack
The official implementation of 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.
pmzzs/standard-readme
A standard style for README files