공개 딥러닝 라이브러리에 대한 보안 취약성 검증
Security Vulnerability Verification for Open Deep Learning Libraries
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
최근 다양한 분야에서 활용중인 딥러닝은 적대적 공격 가능성의 발견으로 위험성이 제기되고 있다. 본 논문에서는딥러닝의 이미지 분류 모델에서 악의적 공격자가 생성한 적대적 샘플에 의해 분류 정확도가 낮아짐을 실험적으로 검증하였다. 대표적인 이미지 샘플인 MNIST데이터 셋을 사용하였으며, 텐서플로우와 파이토치라이브러리를 사용하여만든 오토인코더 분류 모델과 CNN(Convolution neural network)분류 모델에 적대적 샘플을 주입하여 탐지정확도를 측정한다. 적대적 샘플은 MNIST테스트 데이터 셋을 JSMA(Jacobian-based Saliency MapAttack)방법으로 생성한 방법과 FGSM(Fast Gradient Sign Method)방식으로 변형하여 생성하였으며, 분류모델에 주입하여 측정하였을 때 최소 21.82%에서 최대 39.08%만큼 탐지 정확도가 낮아짐을 검증하였다.
Abstract
Deep Learning, which is being used in various fields recently, is being threatened with Adversarial Attack. In this paper,we experimentally verify that the classification accuracy is lowered by adversarial samples generated by malicious attackersin image classification models. We used MNIST dataset and measured the detection accuracy by injecting adversarialsamples into the Autoencoder classification model and the CNN (Convolution neural network) classification model, which arecreated using the Tensorflow library and the Pytorch library. Adversarial samples were generated by transforming MNISTtest dataset with JSMA(Jacobian-based Saliency Map Attack) and FGSM(Fast Gradient Sign Method). When injected into theclassification model, detection accuracy decreased by at least 21.82% up to 39.08%.
참고문헌 (12)
- Papernot, Nicolas, "Practical black-box attacks against machine learning", Proceedings of the 2017ACM on Asia Conference on Computer and Communications Security, pp.506~519 (2017), ACM
- Zhang, Guoming, "Dolphin Attack: Inaudible voice commands", Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security, pp.103~117 (2017), ACM
- Kurakin, Alexey, "Adversarial examples in the physical world", arXiv preprint arXiv : 1607. 02533, ~ (2016)
- Finlayson, "Adversarial Attacks Against Medical Deep Learning Systems", arXiv preprint arXiv : 1804. 05296, ~ (2018)
- Ian J Goodfellow, "Explaining and harnessing adversarial examples", arXiv preprint arXiv:1412.6572, ~ (2014)
- Papernot, Nicolas, "The limitations of deep learning in adversarial settings", Security and Privacy (EuroS&P), 2016 IEEE European Symposium on, pp.372~387 (2016), IEEE
- Papernot, Nicolas, "cleverhans v2. 0.0: an adversarial machine learning library", arXiv preprint arXiv:1610.00768, ~ (2016)
- Vincent, Pascal, "Stacked denoising autoencoders: Learning useful representations in a deep network with a local denoising criterion", Journal of machine learning research, 11, pp.3371~3408 (2010)
- Heckerman, David, "Models and selection criteria for regression and classification", Proceedings of the Thirteenth conference on Uncertainty in artificial intelligence, pp.223~228 (1997), Morgan Kaufmann Publishers Inc
- Krizhevsky, Alex, "Imagenet classification with deep convolutional neural networks", Advances in neural information processing systems, pp.1097~1105 (2012)
- Bottou, Léon, "Comparison of classifier methods: a case study in handwritten digit recognition", Pattern Recognition, 1994. Vol. 2-Conference B: Computer Vision & Image Processing., Proceedings of the 12th IAPR International. Conference on, 2, pp.77~82 (1994), IEEE
- Abadi, Martín, "Tensorflow : a system for large-scale machine learning", OSDI, 16, pp.265~283 (2016)