Attack to Explain Deep Representation

Jalwana, Mohammad A. A. K.; Akhtar, Naveed; Bennamoun, Mohammed; Mian, Ajmal

doi:10.1109/cvpr42600.2020.00956

Cited by 15 publications

(12 citation statements)

References 29 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In essence, this demonstrates a larger perceptual alignment between CapsNet representation and human visual system as compared to CNNs. For reference, perceptual alignment between deep visual models and human vision is also discussed at length in [154], [289]. In [387], the authors proposed to leverage Lightweight Bayesian neural networks for task agnostic detection of adversarial perturbations in inputs using Bayes principle.…”

Section: B Detection For Defensementioning

confidence: 99%

“…Jalwana et al [289] developed a technique to visually reveal the understanding of human-defined semantic concepts by deep learning perceptual models, see Fig. 16.…”

Section: B the Link Between Attacks And Model Interpretationmentioning

confidence: 99%

“…For instance, Xu et al [293] applied group sparsity over the perturbation vector and showed that the resulting perturbations are more interpretable. Nevertheless, this method does not offer interpretability of the model itself like [289], [154].…”

Section: B the Link Between Attacks And Model Interpretationmentioning

confidence: 99%

“…. Jalwana et al[289] proposed an 'attack to explain' which uses perturbations to visualize human-defined semantic concepts as understood by a model. They also utilized their attack to perform low-level vision tasks by attacking robust classifiers.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey

2018

Self Cite

View full text Add to dashboard Cite

Deep learning is at the heart of the current rise of artificial intelligence. In the field of Computer Vision, it has become the workhorse for applications ranging from self-driving cars to surveillance and security. Whereas deep neural networks have demonstrated phenomenal success (often beyond human capabilities) in solving complex problems, recent studies show that they are vulnerable to adversarial attacks in the form of subtle perturbations to inputs that lead a model to predict incorrect outputs. For images, such perturbations are often too small to be perceptible, yet they completely fool the deep learning models. Adversarial attacks pose a serious threat to the success of deep learning in practice. This fact has recently lead to a large influx of contributions in this direction. This article presents the first comprehensive survey on adversarial attacks on deep learning in Computer Vision. We review the works that design adversarial attacks, analyze the existence of such attacks and propose defenses against them. To emphasize that adversarial attacks are possible in practical conditions, we separately review the contributions that evaluate adversarial attacks in the real-world scenarios. Finally, drawing on the reviewed literature, we provide a broader outlook of this research direction.

show abstract

Section: B Detection For Defensementioning

confidence: 99%

“…Jalwana et al [289] developed a technique to visually reveal the understanding of human-defined semantic concepts by deep learning perceptual models, see Fig. 16.…”

Section: B the Link Between Attacks And Model Interpretationmentioning

confidence: 99%

Section: B the Link Between Attacks And Model Interpretationmentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey

2018

Self Cite

View full text Add to dashboard Cite

show abstract

“…However, the presented visual results are far from acceptable. More recently, Jalwana et al [52] demonstrated that attacks can also be a useful tool for model explanation. However, the authors still advocate the need of defense techniques in adversarial settings.…”

Section: Beyond Attacks and Defensesmentioning

confidence: 99%

Orthogonal Deep Models as Defense Against Black-Box Attacks

et al. 2020

Self Cite

View full text Add to dashboard Cite

Deep learning has demonstrated state-of-the-art performance for a variety of challenging computer vision tasks. On one hand, this has enabled deep visual models to pave the way for a plethora of critical applications like disease prognostics and smart surveillance. On the other, deep learning has also been found vulnerable to adversarial attacks, which calls for new techniques to defend deep models against these attacks. Among the attack algorithms, the black-box schemes are of serious practical concern since they only need publicly available knowledge of the targeted model. We carefully analyze the inherent weakness of deep models in black-box settings where the attacker may develop the attack using a model similar to the targeted model. Based on our analysis, we introduce a novel gradient regularization scheme that encourages the internal representation of a deep model to be orthogonal to another, even if the architectures of the two models are similar. Our unique constraint allows a model to concomitantly endeavour for higher accuracy while maintaining near orthogonal alignment of gradients with respect to a reference model. Detailed empirical study verifies that controlled misalignment of gradients under our orthogonality objective significantly boosts a model's robustness against transferable black-box adversarial attacks. In comparison to regular models, the orthogonal models are significantly more robust to a range of l p norm bounded perturbations. We verify the effectiveness of our technique on a variety of large-scale models. INDEX TERMSDeep learning, adversarial examples, adversarial perturbations, orthogonal models, robust deep learning.

show abstract