Modeling Adversarial Noise for Adversarial Training

Authors: Dawei Zhou, Nannan Wang, Bo Han, Tongliang Liu

ICML 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 4. Experiments In this section, we first introduce the experiment setup in Section 4.1. Then, we evaluate the effectiveness of our defense method against representative and commonly used L∞ norm and L2-norm adversarial attacks in Section 4.2. In addition, we conduct ablation studies in Section 4.3.
Researcher Affiliation Academia 1ISN Lab, School of Telecommunications Engineering, Xidian University, dwzhou.xidian@gmail.com, nnwang@xidian.edu.cn 2TML Lab, Sydney AI Centre, The University of Sydney 3Department of Computer Science, Hong Kong Baptist University, bhanml@comp.hkbu.edu.hk.
Pseudocode Yes Algorithm 1 Training the defense model based on Modeling Adversarial Noise (MAN).
Open Source Code Yes The code is available at https://github.com/dwDavidxd/MAN.
Open Datasets Yes Datasets. We verify the effective of our defense method on two popular benchmark datasets, i.e., CIFAR-10 (Krizhevsky et al., 2009) and Tiny-Image Net (Wu et al., 2017).
Dataset Splits Yes CIFAR-10 has 10 classes of images including 50,000 training images and 10,000 test images. Tiny-Image Net has 200 classes of images including 100,000 training images, 10,000 validation images and 10,000 test images.
Hardware Specification No No specific hardware details such as GPU/CPU models, processor types, or memory amounts used for running experiments are mentioned. The text only refers to model architectures like ResNet-18 and VggNet-19.
Software Dependencies No The paper mentions training models using SGD, but does not provide specific software names with version numbers for libraries, frameworks, or languages used for implementation.
Experiment Setup Yes For all baselines and our defense method, we use the L∞-norm non-target PGD-10 (i.e., PGD with iteration number of 10) with random start and step size ϵ/4 to craft adversarial training data. The perturbation budget ϵ is set to 8/255 for both CIFAR-10 and Tiny-Image Net. All the defense models are trained using SGD with momentum 0.9 and an initial learning rate of 0.1. The weight decay is 2 × 10−4 for CIFAR-10, and is 5 × 10−4 for Tiny-Image Net. The batch-size is set as 1024 to reduce time cost. The epoch number is set to 100. The learning rate is divided by 10 at the 75-th and 90-th epoch.