| 引用本文: | 凌祥,周伯霖,王时予,罗天悦,尹鹏,吴春明,王滨,吴敬征.RMDroid: 基于多模态融合学习的安卓恶意软件鲁棒检测.软件学报,2026,37(4):1715-1739 |
| |
|
| |
|
|
| 本文已被：浏览 904次 下载 2027次 |
码上扫一扫！ |
|
|
| RMDroid: 基于多模态融合学习的安卓恶意软件鲁棒检测 |
|
凌祥1,2, 周伯霖1,2, 王时予1,2, 罗天悦1, 尹鹏2,3, 吴春明4, 王滨5, 吴敬征1,2
|
|
1.中国科学院 软件研究所, 北京 100190;2.中国科学院大学, 北京 100049;3.军工保密资格审查认证中心, 北京 100089;4.浙江大学 计算机科学与技术学院, 浙江 杭州 310027;5.杭州海康威视网络与信息安全实验室, 浙江 杭州 310052
|
|
| 摘要: |
| 随着人工智能技术的蓬勃发展和广泛应用, 越来越多的恶意软件检测方法和工具利用深度学习的强大学习能力来检测安卓平台上新出现的恶意软件. 然而, 深度学习模型已经被证明容易受到对抗攻击的威胁. 与此同时, 攻击者已经开始提出多种针对安卓恶意软件检测方法的对抗攻击方法, 即生成对抗性安卓恶意软件, 从而达到绕过恶意软件检测的目的. 现有安卓恶意软件检测方法容易受到对抗攻击威胁的主要原因在于, 这些恶意软件检测方法都建立在单一模态特征之上, 而以单一模态存在的特征却很容易被攻击者恶意性地操控. 因此, 为了提高当前安卓恶意软件检测方法可以抵御对抗攻击的鲁棒性, 提出一种基于多模态融合学习的安卓恶意软件鲁棒检测方法RMDroid, 可以在不影响针对一般性安卓恶意软件检测准确性的基础上, 显著提高其抵御对抗攻击的鲁棒性. 具体而言, RMDroid首先会从待测安卓软件中提取多种模态的特征信息, 然后分别利用相应的深度学习模型学习表征相应模态深层语义信息的特征向量, 最后利用异类识别网络降低甚至消除多模态特征中受到对抗攻击干扰的模态特征对最终恶意软件预测的影响, 从而提高其抵御对抗攻击的鲁棒性. 实验结果表明, 所提出的RMDroid在5项有效性指标和1项鲁棒性指标上均优于所有基线检测方法. 特别的, 在误报率FPR相同的情况下, RMDroid的检出率TPR比最好的基线检测方法的TPR高出10%以上; 并且针对最先进的HRAT攻击, RMDroid的鲁棒性值高达96%以上, 显著高于MaMaDroid和MalScan基线检测方法的鲁棒性值. |
| 关键词: 安卓恶意软件 鲁棒检测 对抗攻击 多模态学习 融合学习 |
| DOI：10.13328/j.cnki.jos.007499 |
| 分类号:TP311 |
| 基金项目:国家自然科学基金(62202457); 中国博士后科学基金(2022M713253); 国防基础科研计划(JCKY2021906B002) |
|
| RMDroid: Android Malware Robust Detection Based on Multi-modal Fusion Learning |
|
LING Xiang1,2, ZHOU Bo-Lin1,2, WANG Shi-Yu1,2, LUO Tian-Yue1, YIN Peng2,3, WU Chun-Ming4, WANG Bin5, WU Jing-Zheng1,2
|
|
1.Institute of Software, Chinese Academy of Sciences, Beijing 100190, China;2.University of Chinese Academy of Sciences, Beijing 100049, China;3.Defense Industry Secrecy Examination and Certification Center, Beijing 100089, China;4.College of Computer Science and Technology, Zhejiang University, Hangzhou 310027, China;5.Network and Information Security Laboratory of Hangzhou Hikvision Digital Technology Co. Ltd., Hangzhou 310052, China
|
| Abstract: |
| With the booming development and wide application of artificial intelligence, more and more deep learning-based Android malware detection methods and tools have been developed to detect newly emerged Android malware. However, deep learning models have been extensively proven to be vulnerable to adversarial attacks. Meanwhile, attacker shave started to propose adversarial attacks against Android malware detection methods to generate adversarial Android malware that can bypass detection. This study argues that the main reason current Android malware detection methods are vulnerable to such adversarial attacks is that these detectors are mostly built on single-modal features, which can be easily manipulated by attackers. Therefore, to improve the robustness of Android malware detection against adversarial attacks, the study proposes a robust Android malware detection method based on multi-modal fusion learning, namely RMDroid. RMDroid improves robustness in identifying adversarial malware without sacrificing accuracy in general Android malware detection. Specifically, RMDroid first extracts feature information from different modalities of Android APPs and then uses the corresponding deep learning models to sufficiently learn feature vectors that characterize the deep semantics of each modality. Finally, an odd-one-out network is employed to reduce or even eliminate the influence of interfered modal features on the final malware prediction, thus improving robustness against adversarial attacks. The experimental results show that RMDroid achieves higher performance across five effectiveness metrics and one robustness metric compared to all baseline detection methods. In particular, given the same FPR, the TPR value of RMDroid is more than 10% higher than that of the best baseline detection method. In the case of the state-of-the-art adversarial attack of HRAT, RMDroid achieves over 96% in robustness, which is significantly higher than the robustness of both MaMaDroid and MalScan. |
| Key words: Android malware robust detection adversarial attack multi-modal learning fusion learning |
|
|
|
|
