악성코드 변종 분석을 위한 AI 모델의 Robust 수준 측정 및 개선 연구
A Study on Robustness Evaluation and Improvement of AI Model for Malware Variation Analysis
0
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
오늘날 AI(Artificial Intelligence) 기술은 악성코드 분야를 비롯하여 다양한 분야에서 광범위하게 연구되고 있다. 중요한 의사결정 및 자원을 보호하는 역할에 AI 시스템을 도입하기 위해서는 신뢰할 수 있는 AI 모델이어야 한다. 학습 데이터셋에 의존적인 AI 모델은 새로운 공격에 대해서도 견고한지 확인이 필요하다. 공격자는 악성코드를 새로 생성하기보단, 기존에 탐지되었던 악성코드의 변종을 대량 생산하여 공격에 성공하는 악성코드를 탐색다. AI 모델의 Misclassification을 유도하는 Adversarial attack과 같이 대부분의 공격은 기존 공격에 약간에 변형을 가해 만든 공격들이다. 이러한 변종에도 대응 가능한 Robust한 모델이 필요하며, AI 평가지표로 많이 사용되는 Accuracy, Recall 등으로는 모델의 Robustness 수준을 측정할 수 없다. 본 논문에서는 Adversarial attack 중 하나인 C&W attack을 기반으로 Adversarial sample을 생성하여 Robustness 수준을 측정하고 Adversarial training 을 통해 Robustness 수준을 개선하는 방법을 실험한다. 본 연구의 악성코드 데이터셋 기반 실험을 통해 악성코드 분야에서 해당 제안 방법의 한계 및 가능성을 확인하였다.
Abstract
Today, AI(Artificial Intelligence) technology is being extensively researched in various fields, including the field of malware detection. To introduce AI systems into roles that protect important decisions and resources, it must be a reliable AI model. AI model that dependent on training dataset should be verified to be robust against new attacks. Rather than generating new malware detection, attackers find malware detection that succeed in attacking by mass-producing strains of previously detected malware detection. Most of the attacks, such as adversarial attacks, that lead to misclassification of AI models, are made by slightly modifying past attacks. Robust models that can be defended against these variants is needed, and the Robustness level of the model cannot be evaluated with accuracy and recall, which are widely used as AI evaluation indicators. In this paper, we experiment a framework to evaluate robustness level by generating an adversarial sample based on one of the adversarial attacks, C&W attack, and to improve robustness level through adversarial training. Through experiments based on malware dataset in this study, the limitations and possibilities of the proposed method in the field of malware detection were confirmed.
참고문헌 (16)
- Hamon, Ronan, "Robustness and explainability of artificial intelligence" (2020), Publications Office of the European Union
- Diro, A. A., "Distributed attack detection scheme using deep learning approach for Internet of Things", Future Generation Computer Systems, 82, pp.761~768 (2018)
- Shafique, Muhammad, "Robust machine learning systems : Challenges, current trends, perspectives, and the road ahead", IEEE Design & Test, 37(2), pp.30~57 (2020)
- Goodfellow, Ian J., "Explaining and harnessing adversarial examples", arXiv preprint arXiv:1412.6572, ~ (2014)
- Biggio, Battista, "Wild patterns: Ten years after the rise of adversarial machine learning", Pattern Recognition, 84, pp.317~331 (2018)
- Carlini, Nicholas, "On evaluating adversarial robustness", arXiv preprint arXiv:1902.06705, ~ (2019)
- 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)
- Papernot, Nicolas, "The limitations of deep learning in adversarial settings", 2016 IEEE European symposium on security and privacy (EuroS&P). IEEE, pp.372~387 (2016)
- Chen, Pin-Yu, "Zoo: Zeroth order optimization based black-box attacks to deep neural networks without training substitute models", Proceedings of the 10th ACM workshop on artificial intelligence and security, pp.15~26 (2017)
- Carlini, Nicholas, "Towards evaluating the robustness of neural networks", 2017 ieee symposium on security and privacy (sp). Ieee, pp.39~57 (2017)
- Chang, Chih-Ling, "Evaluating robustness of ai models against adversarial attacks", Proceedings of the 1st ACM Workshop on Security and Privacy on Artificial Intelligence, pp.47~54 (2020)
- Nicolae, Maria-Irina, "Adversarial Robustness Toolbox v1. 0.0", arXiv preprint arXiv:1807.01069, ~ (2018)
- dRauber, Jonas, "Foolbox: A python toolbox to benchmark the robustness of machine learning models", arXiv preprint arXiv:1707.04131, ~ (2017)
- Papernot, Nicolas, "Technical report on the cleverhans v2. 1.0adversarial examples library", arXiv preprint arXiv:1610.00768, ~ (2016)
- Berghoff, Christian, "Robustness testing of ai systems: a case study for traffic sign recognition", IFIP International Conference on Artificial Intelligence Applications and Innovations, pp.256~267 (2021), Springer
- Hartl, Alexander, "Explainability and adversarial robustness for rnns", 2020 IEEE Sixth International Conference on Big Data Computing Service and Applications (BigDataService). IEEE, pp.148~156 (2020)