적대적 공격에 견고한 Perceptual Ad-Blocker 기법
Perceptual Ad-Blocker Design For Adversarial Attack
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
Perceptual Ad-Blocking은 인공지능 기반의 광고 이미지 분류 모델을 이용하여 온라인 광고를 탐지하는 새로운 광고 차단 기법이다. 이러한 Perceptual Ad-Blocking은 최근 이미지 분류 모델이 이미지를 틀리게 분류하게끔 이미지에 노이즈를 추가하는 적대적 예제(adversarial example)를 이용한 적대적 공격(adversarialattack)에 취약하다는 연구 결과가 제시된 바 있다. 본 논문에서는 다양한 적대적 예제를 통해 기존 PerceptualAd-Blocking 기법의 취약점을 증명하고, MNIST, CIFAR-10 등의 데이터 셋에서 성공적인 방어를 수행한Defense-GAN과 MagNet이 광고 이미지에도 효과적으로 작용함을 보인다. 이를 통해 Defense-GAN과MagNet 기법을 이용해 적대적 공격에 견고한 새로운 광고 이미지 분류 모델을 제시한다. 기존 다양한 적대적 공격기법을 이용한 실험 결과에 따르면, 본 논문에서 제안하는 기법은 적대적 공격에 견고한 이미지 분류 기술을 통해공격 이전의 이미지 분류 모델의 정확도와 성능을 확보할 수 있으며, 더 나아가 방어 기법의 세부사항을 아는 공격자의 화이트박스 공격(White-box attack)에도 일정 수준 방어가 가능함을 보였다.
Abstract
Perceptual Ad-Blocking is a new advertising blocking technique that detects online advertising by using an artificialintelligence-based advertising image classification model. A recent study has shown that these Perceptual Ad-Blocking modelsare vulnerable to adversarial attacks using adversarial examples to add noise to images that cause them to be misclassified. In this paper, we prove that existing perceptual Ad-Blocking technique has a weakness for several adversarial example andthat Defense-GAN and MagNet who performed well for MNIST dataset and CIFAR-10 dataset are good to advertisingdataset. Through this, using Defense-GAN and MagNet techniques, it presents a robust new advertising image classificationmodel for adversarial attacks. According to the results of experiments using various existing adversarial attack techniques, thetechniques proposed in this paper were able to secure the accuracy and performance through the robust image classificationtechniques, and furthermore, they were able to defend a certain level against white-box attacks by attackers who knew thedetails of defense techniques.
참고문헌 (20)
- C. Szegedy, "Intriguing properties of neural networks", arXiv preprint arXiv:1312.6199, ~ (2013)
- I. J. Goodfellow, "Explaining and harnessing adversarial examples", arXiv preprint arXiv:1412.6572, ~ (2014)
- F. Tramèr, "Adversarial: perceptual ad blocking meets adversarial machine learning", Proceedings of the 2019 ACM SIGSAC Conference on Computer and Communications Secur, pp.2005~2021 (2019)
- P. Samangouei, "Defense-gan: protecting classifiers against adversarial attacks using generative models", arXiv preprint arXiv:1805.06605, ~ (2018)
- D. Meng, "Magnet: a two-pronged defense against adversarial examples", Proceedings of the 2017 ACM SIGSAC conference on computer and communications security, pp.135~147 (2017)
- I. Goodfellow, "Generative adversarial nets", Advances in neural information processing systems, pp.2014~2680 (2014)
- Z. Hussain, "Automatic understanding of image and video advertisements", Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp.1705~1715 (2017)
- Z. A. Din, "Percival: making in-browser perceptual ad blocking practical with deep learning", 2020 USENIX Annual Technical Conference, pp.387~400 (2020)
- "Broadcasting and Communica tions Advertising Expenses Survey" (2020)[URL]
- A. Kurakin, "Adversarial examples in the physical world", arXiv preprint arXiv:1607.02533, ~ (2016)
- A. Madry, "Towards deep learning models resistant to adversarial attacks", arXiv preprint arXiv:1706.06083, ~ (2017)
- S. M. Moosavi-Dezfooli, "Deepfool: a simple and accurate method to fool deep neural networks", Proceedings of the IEEE conference on computer vision and pattern recognition, pp.2574~2582 (2016)
- K. He, "Deep residual learning for image recognition", Proceedings of the IEEE conference on computer vision and pattern recognition, pp.770~778 (2016)
- A. Athalye, "Obfuscated gradients give a false sense of security: circumventing defenses to adversarial examples", International Conference on Machine Learning, pp.274~283 (2018)
- S. Qiu, "Review of artificial intelligence adversarial attack and defense technologies", Applied Sciences, 9(5), pp.909~ (2019)
- N. Carlini, "Towards evaluating the robustness of neural networks", IEEE symposium on security and privacy, pp.39~57 (2017)
- Learning multiple layers of features from tiny images, "A. Krizhevsky", University of Toronto, ~ (2009)
- "The mnist database of handwritten digits" (1998)[URL]
- H. Xiao, "Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms", arXiv preprint arXiv:1708.07747, ~ (2017)
- M. Arjovsky, "Wasserstein generative adversarial networks", Proceedings of the 34th International Conference on Machine Learning, 70, pp.214~223 (2017)