Improving Adversarial Robustness via Feature Pattern Consistency Constraint

Authors: Jiacong Hu, Jingwen Ye, Zunlei Feng, Jiazhen Yang, Shunyu Liu, Xiaotian Yu, Lingxiang Jia, Mingli Song

IJCAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experiments demonstrate that the FPCC method empowers latent features to uphold correct feature patterns even in the face of adversarial examples, resulting in inherent adversarial robustness surpassing state-of-the-art models.
Researcher Affiliation Academia Jiacong Hu1,3,4 , Jingwen Ye5 , Zunlei Feng2,3,4, , Jiazhen Yang2 , Shunyu Liu1 , Xiaotian Yu1 , Lingxiang Jia1 and Mingli Song1,3,4 1College of Computer Science and Technology, Zhejiang University 2School of Software Technology, Zhejiang University 3State Key Laboratory of Blockchain and Security, Zhejiang University 4Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security 5National University of Singapore
Pseudocode Yes Algorithm 1 Training Model with PRO
Open Source Code No The paper does not contain any explicit statement about releasing source code or a link to a code repository.
Open Datasets Yes We conducted experiments using two widely utilized datasets: CIFAR-10 [Krizhevsky, 2009] and CIFAR-100 [Krizhevsky, 2009].
Dataset Splits No The paper mentions using CIFAR-10 and CIFAR-100 datasets but does not explicitly provide specific training, validation, or test dataset splits or a reference to how these splits were defined.
Hardware Specification No The paper does not provide specific details about the hardware used for running experiments, such as GPU models, CPU models, or cloud computing specifications.
Software Dependencies No The paper does not provide specific ancillary software details, such as library names with version numbers, needed to replicate the experiment.
Experiment Setup Yes The total loss in PRO is given as: L = LCE + λ X l N L(l) P T = i=1 LCE,i + λ l N L(l) P T,i i=1 log( exp(zi[yi]) PK k=1 exp(zi[k]) ) + λ l N ||p(l) i d(l) yi ||1, (11) where λ is a balancing factor for the two losses. The entire PRO process is end-to-end, as shown in Algorithm 1.