Adversarial Contrastive Learning by Permuting Cluster Assignments

Wahed, Muntasir; Tabassum, Afrina; Lourentzou, Ismini

doi:10.48550/arxiv.2204.10314

Cited by 2 publications

(3 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Adversarial vulnerability under standard training: In agreement with [5], [7], [8], the results obtained from scenario ST-ST show the adversarial vulnerability of the studied contrastive learning schemes. We also observed that semisupervised learning schemes SL-CL or SCL-CL achieve better robust performance than CL scheme.…”

Section: Discussionsupporting

confidence: 85%

“…Kim et al [5] were the first to utilize the contrastive loss to generate adversarial examples without any label for robustifying of SimCLR [6] framework. Moshavash et al [7] and Wahed et al [8] have applied the same technique on Momentum Contrast (MOCO) [9] and Swapping Assignments between Views (SwAV) [10], respectively. Fan et al [11] introduced an additional regularization term in contrastive loss to enhance cross-task robustness transferability.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

On the Role of Contrastive Representation Learning in Adversarial Robustness: An Empirical Study

Ghofrani¹,

Yaghouti²,

Jamshidi³

2023

Preprint

View full text Add to dashboard Cite

Self-supervised contrastive learning has solved one of the significant obstacles in deep learning by alleviating the annotation cost. This advantage comes with the price of false negative-pair selection without any label information. Supervised contrastive learning has emerged as an extension of contrastive learning to eliminate this issue. However, aside from accuracy, there is a lack of understanding about the impacts of adversarial training on the representations learned by these learning schemes. In this work, we utilize supervised learning as a baseline to comprehensively study the robustness of contrastive and supervised contrastive learning under different adversarial training scenarios. Then, we begin by looking at how adversarial training affects the learned representations in hidden layers, discovering more redundant representations between layers of the model. Our results on CIFAR-10 and CIFAR-100 image classification benchmarks demonstrate that this redundancy is highly reduced by adversarial fine-tuning applied to the contrastive learning scheme, leading to more robust representations. However, adversarial fine-tuning is not very effective for supervised contrastive learning and supervised learning schemes. Our code is released at https://github.com/softsys4ai/CL-Robustness.

show abstract

Section: Discussionsupporting

confidence: 85%

Section: Introductionmentioning

confidence: 99%

On the Role of Contrastive Representation Learning in Adversarial Robustness: An Empirical Study

Ghofrani¹,

Yaghouti²,

Jamshidi³

2023

Preprint

View full text Add to dashboard Cite

show abstract

“…All other settings are the same as original papers except for some hyperparameter tuning. Our methods are also compatible with some recent work like SwARo [46] and CLAF [38], by modeling the asymmetry between clean and adversarial views as aforementioned.…”

Section: Methodsmentioning

confidence: 70%

Adversarial Contrastive Learning via Asymmetric InfoNCE

Yu¹,

Lou²,

Zhan³

et al. 2022

Preprint

View full text Add to dashboard Cite

Contrastive learning (CL) has recently been applied to adversarial learning tasks. Such practice considers adversarial samples as additional positive views of an instance, and by maximizing their agreements with each other, yields better adversarial robustness. However, this mechanism can be potentially flawed, since adversarial perturbations may cause instance-level identity confusion, which can impede CL performance by pulling together different instances with separate identities. To address this issue, we propose to treat adversarial samples unequally when contrasted, with an asymmetric InfoNCE objective (A-InfoNCE ) that allows discriminating considerations of adversarial samples. Specifically, adversaries are viewed as inferior positives that induce weaker learning signals, or as hard negatives exhibiting higher contrast to other negative samples. In the asymmetric fashion, the adverse impacts of conflicting objectives between CL and adversarial learning can be effectively mitigated. Experiments show that our approach consistently outperforms existing Adversarial CL methods across different finetuning schemes. The proposed A-InfoNCE is also a generic form that can be readily extended to other CL methods. Code is available at https://github.com/yqy2001/A-InfoNCE.

show abstract

Adversarial Contrastive Learning by Permuting Cluster Assignments

Cited by 2 publications

References 17 publications

On the Role of Contrastive Representation Learning in Adversarial Robustness: An Empirical Study

On the Role of Contrastive Representation Learning in Adversarial Robustness: An Empirical Study

Adversarial Contrastive Learning via Asymmetric InfoNCE

Contact Info

Product

Resources

About