Improving adversarial robustness by a coupling rejection strategy

Last update: Jan 06, 2023

Related tags

Deep Learning Rectified-Rejection

Overview

Adversarial Training with Rectified Rejection

The code for the paper Adversarial Training with Rectified Rejection.

Environment settings and libraries we used in our experiments

This project is tested under the following environment settings:

OS: Ubuntu 18.04.4
GPU: Geforce 2080 Ti or Tesla P100
Cuda: 10.1, Cudnn: v7.6
Python: 3.6
PyTorch: >= 1.6.0
Torchvision: >= 0.6.0

Acknowledgement

The codes are modifed based on Rice et al. 2020, and the model architectures are implemented by pytorch-cifar.

Training Commands

Below we provide running commands training the models with the RR module, taking the setting of PGD-AT + RR (ResNet-18) as an example:

python train_cifar.py --model_name PreActResNet18_twobranch_DenseV1 --attack pgd --lr-schedule piecewise \
                                              --epochs 110 --epsilon 8 \
                                              --attack-iters 10 --pgd-alpha 2 \
                                              --fname auto \
                                              --batch-size 128 \
                                              --adaptivetrain --adaptivetrainlambda 1.0 \
                                              --weight_decay 5e-4 \
                                              --twobranch --useBN \
                                              --selfreweightCalibrate \
                                              --dataset 'CIFAR-10' \
                                              --ATframework 'PGDAT' \
                                              --SGconfidenceW

The FLAG --model_name can be PreActResNet18_twobranch_DenseV1 (ResNet-18) or WideResNet_twobranch_DenseV1 (WRN-34-10). For alternating different AT frameworks, we can set the FLAG --ATframework to be one of PGDAT, TRADES, CCAT.

Evaluation Commands

Below we provide running commands for evaluations.

Evaluating under the PGD attacks

The trained model is saved at trained_models/model_path, where the specific name of model_path is automatically generated during training. The command for evaluating under PGD attacks is:

python eval_cifar.py --model_name PreActResNet18_twobranch_DenseV1 --evalset test --norm l_inf --epsilon 8 \
                                              --attack-iters 1000 --pgd-alpha 2 \
                                              --fname trained_models/model_path \
                                              --load_epoch -1 \
                                              --dataset 'CIFAR-10' \
                                              --twobranch --useBN \
                                              --selfreweightCalibrate

Evaluating under the adaptive CW attacks

The parameter FLAGs --binary_search_steps, --CW_iter, --CW_confidence can be changed, where --detectmetric indicates the rejector that needs to be adaptively evaded.

python eval_cifar_CW.py --model_name PreActResNet18_twobranch_DenseV1 --evalset adaptiveCWtest \
                                              --fname trained_models/model_path \
                                              --load_epoch -1 --seed 2020 \
                                              --binary_search_steps 9 --CW_iter 100 --CW_confidence 0 \
                                              --threatmodel linf --reportmodel linf \
                                              --twobranch --useBN \
                                              --selfreweightCalibrate \
                                              --detectmetric 'RR' \
                                              --dataset 'CIFAR-10'

Evaluating under multi-target and GAMA attacks

The running command for evaluating under multi-target attacks is activated by the FLAG --evalonMultitarget as:

python eval_cifar.py --model_name PreActResNet18_twobranch_DenseV1 --evalset test --norm l_inf --epsilon 8 \
                                              --attack-iters 100 --pgd-alpha 2 \
                                              --fname trained_models/model_path \
                                              --load_epoch -1 \
                                              --dataset 'CIFAR-10' \
                                              --twobranch --useBN \
                                              --selfreweightCalibrate \
                                              --evalonMultitarget --restarts 1

The running command for evaluating under GAMA attacks is activated by the FLAG --evalonGAMA_PGD or --evalonGAMA_FW as:

python eval_cifar.py --model_name PreActResNet18_twobranch_DenseV1 --evalset test --norm l_inf --epsilon 8 \
                                              --attack-iters 100 --pgd-alpha 2 \
                                              --fname trained_models/model_path \
                                              --load_epoch -1 \
                                              --dataset 'CIFAR-10' \
                                              --twobranch --useBN \
                                              --selfreweightCalibrate \
                                              --evalonGAMA_FW

Evaluating under CIFAR-10-C

The running command for evaluating on common corruptions in CIFAR-10-C is:

python eval_cifar_CIFAR10-C.py --model_name PreActResNet18_twobranch_DenseV1 \
                                              --fname trained_models/model_path \
                                              --load_epoch -1 \
                                              --dataset 'CIFAR-10' \
                                              --twobranch --useBN \
                                              --selfreweightCalibrate

Improving adversarial robustness by a coupling rejection strategy

Related tags

Overview

Adversarial Training with Rectified Rejection

Environment settings and libraries we used in our experiments

Acknowledgement

Training Commands

Evaluation Commands

Evaluating under the PGD attacks

Evaluating under the adaptive CW attacks

Evaluating under multi-target and GAMA attacks

Evaluating under CIFAR-10-C

Owner

Tianyu Pang

Test-Time Personalization with a Transformer for Human Pose Estimation, NeurIPS 2021

CLIP + VQGAN / PixelDraw

A Python implementation of global optimization with gaussian processes.

Robust fine-tuning of zero-shot models

CZU-MHAD: A multimodal dataset for human action recognition utilizing a depth camera and 10 wearable inertial sensors

Official pytorch implementation of "Scaling-up Disentanglement for Image Translation", ICCV 2021.

Generative code template for PixelBeasts 10k NFT project.

MPRNet-Cloud-removal: Progressive cloud removal

Few-Shot Graph Learning for Molecular Property Prediction

Image-to-Image Translation in PyTorch

Code for "Causal autoregressive flows" - AISTATS, 2021

Libraries, tools and tasks created and used at DeepMind Robotics.

Code for our ICCV 2021 Paper "OadTR: Online Action Detection with Transformers".

Codes and Data Processing Files for our paper.

Ensemble Knowledge Guided Sub-network Search and Fine-tuning for Filter Pruning

Official PyTorch Implementation for "Recurrent Video Deblurring with Blur-Invariant Motion Estimation and Pixel Volumes"

Code for the paper Relation Prediction as an Auxiliary Training Objective for Improving Multi-Relational Graph Representations (AKBC 2021).

RodoSol-ALPR Dataset

Code for the AAAI 2022 paper "Zero-Shot Cross-Lingual Machine Reading Comprehension via Inter-Sentence Dependency Graph".

a project for 3D multi-object tracking