qi-smile/adba

Unofficial PyTorch implementation of Anti Distillation Backdoor Attacks(MM '21)

Anti Distillation Backdoor Attacks

This repository is an unofficial PyTorch implementation of:

Anti Distillation Backdoor Attacks(MM '21)

which is closed source.:fu:

Requirements

• This codebase is written for python3 (used python 3.6 while implementing).
• We use Pytorch version of 1.8.2, 11.4 CUDA version.
• To install necessary python packages,

conda env create -f environment.yml
pip install -r requirements.txt

How to Run Codes?

Local Training

python3 main.py --type=pretrain  --lr=0.001 --model=res --dataset=cifar10 --partition=iid --seed=1 --local_ep=200  --alpha_backdoor=0.4 --txtpath=./saved/adba_result.txt

Global Distillation

python3 main.py --type=distillation  --lr=0.001 --model=res --dataset=cifar10 --partition=iid --seed=1 --epochs=200  --alpha_backdoor=0.4 --txtpath=./saved/adba_result.txt

Acknowledgement

This repository is based on the code of DENSE(NeurIPS '22), and I also learn a lot from it when coding.🥰