A Survey of Machine Learning-Based Malware Detection Methods

doi:10.3969/j.issn.1671-1122.2026.04.002

Abstract

Abstract:

With the escalating threats in cyberspace, the volume and complexity of malware have grown explosively. Machine learning, leveraging its powerful feature extraction capabilities, has been widely applied in malware detection tasks. This paper reviewed recent advances in machine learning-based malware detection techniques. First, it introduced the definition of malware and the detection framework. Then, it comprehensively reviewed the applications of traditional machine learning, deep learning, and graph representation learning in malware detection. Furthermore, a comparative analysis of these three categories of machine learning methods was conducted. Finally, the current technical bottlenecks were summarized, and future research directions were proposed.

Key words: malware detection, machine learning, deep learning, graph representation learning

CLC Number:

TP309

LI Hailong, ZHANG Yunhao, SHEN Xieyang, XING Yuhang, CUI Zhian. A Survey of Machine Learning-Based Malware Detection Methods[J]. Netinfo Security, 2026, 26(4): 521-541.

Figures/Tables 13

References 89

[1]	YE Yanfang, LI Tao, ADJEROH D, et al. A Survey on Malware Detection Using Data Mining Techniques[J]. ACM Computing Surveys (CSUR), 2017, 50(3): 1-40.
[2]	ANTON K. Mobile Malware Evolution in 2024[EB/OL].(2025-03-03)[2025-08-18]. https://securelist.com/mobile-threat-report-2024/115494.
[3]	SHIJO P V, SALIM A. Integrated Static and Dynamic Analysis for Malware Detection[J]. Procedia Computer Science, 2015, 46: 804-811.
[4]	KINABLE J, KOSTAKIS O. Malware Classification Based on Call Graph Clustering[J]. Journal in Computer Virology, 2011, 7(4): 233-245.
[5]	OR-MEIR O, NISSIM N, ELOVICI Y, et al. Dynamic Malware Analysis in the Modern Era—A State of the Art Survey[J]. ACM Computing Surveys (CSUR), 2019, 52(5): 1-48.
[6]	SAEED I A, SELAMAT A, ABUAGOUB A M A. A Survey on Malware and Malware Detection Systems[EB/OL].(2013-04-23)[2025-08-18]. https://www.researchgate.net/profile/Imtithal-Saeed/publication/272238656_A_Survey_on_Malwares_and_Malware_Detection_Systems/links/566284c608ae192bbf8cf1a5/A-Survey-on-Malwares-and-Malware-Detection-Systems.pdf.
[7]	KRAMER S, BRADFIELD J C. A General Definition of Malware[J]. Journal in Computer Virology, 2010, 6(2): 105-114.
[8]	MARION J Y. Ransomware: Extortion is My Business[J]. Communications of the ACM, 2025, 68(5): 36-47.
[9]	WANG Siyu, WU Haiyong, LU Ning, et al. Atsdetector: An Android Trojan Spyware Detection Approach with Multi-Features[EB/OL].(2025-03)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S016740482400525X.
[10]	SERAJ S, PAVLIDIS M, TROVATI M, et al. Maddroid: Malicious Adware Detection in Android Using Deep Learning[J]. Journal of Cyber Security Technology, 2024, 8(3): 163-190.
[11]	YANG Wei, KONG Deguang, XIE Tao, et al. Malware Detection in Adversarial Settings: Exploiting Feature Evolutions and Confusions in Android Apps[C]// Springer.The 33rd Annual Computer Security Applications Conference. Heidelberg: Springer, 2017: 288-302.
[12]	LI Xun, LIU Lei, LIU Yuzhou, et al. Multimodal Fusion for Android Malware Detection Based on Large Pre-Trained Models[EB/OL].(2025-04-03)[2025-08-18]. https://ieeexplore.ieee.org/abstract/document/10948385/.
[13]	ANAND A, SINGH J P, SINGH A K. Smali Code-Based Deep Learning Model for Android Malware Detection[J]. The Journal of Supercomputing, 2025, 81(4): 1-31.
[14]	OU Fan, XU Jian. S3Feature:A Static Sensitive Subgraph-Based Feature for Android Malware Detection[EB/OL].(2022-02-16)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404821003370.
[15]	HUANG Hong, HUANG Weitao, ZHOU Yinghang, et al. Fedroid: A Lightweight and Interpretable Machine Learning-Based Android Malware Detection System[EB/OL].(2025-02-25)[2025-08-18]. https://link.springer.com/article/10.1007/s10586-024-04865-x.
[16]	JEON J, JEONG B, BAEK S, et al. Static Multi Feature-Based Malware Detection Using Multi SPP-Net in Smart Iot Environments[J]. IEEE Transactions on Information Forensics and Security, 2024, 19: 2487-2500.
[17]	KIRAZ O, DOGRU I A. Visualising Static Features and Classifying Android Malware Using a Convolutional Neural Network Approach[EB/OL].(2024-04-24)[2025-08-18]. https://www.mdpi.com/2076-3417/14/11/4772.
[18]	CUI Yuning, SUN Yi, LIN Zhaowen. Droidhook: A Novel API-Hook Based Android Malware Dynamic Analysis Sandbox[EB/OL].(2023-02-24)[2025-08-18]. https://link.springer.com/article/10.1007/s10515-023-00378-w.
[19]	AGGARWAL S, TROIA F D. Malware Classification Using Dynamically Extracted API Call Embeddings[EB/OL].(2024-05-20)[2025-08-18]. https://www.mdpi.com/2076-3417/14/13/5731.
[20]	YAN Pei, TAN Shunquan, WANG Miaohui, et al. Enhanced Dynamic Analysis for Malware Detection with Gradient Attack[EB/OL].(2024-10-07)[2025-08-18]. https://ieeexplore.ieee.org/abstract/document/10706706/.
[21]	THAKUR P, KANSAL V, RISHIWAL V. Hybrid Deep Learning Approach Based on LSTM and CNN for Malware Detection[J]. Wireless Personal Communications, 2024, 136(3): 1879-1901.
[22]	TRIZNA D, DEMETRIO L, BIGGIO B, et al. Nebula: Self-Attention for Dynamic Malware Analysis[J]. IEEE Transactions on Information Forensics and Security, 2024, 19: 6155-6167.
[23]	ZHANG Sanfeng, SU Heng, LIU Hongxian, et al. MPDroid: A Multimodal Pre-Training Android Malware Detection Method with Static and Dynamic Features[EB/OL].(2025-03)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404824005686.
[24]	SONG Yafei, ZHANG Dandan, WANG Jian, et al. Application of Deep Learning in Malware Detection: A Review[EB/OL].(2025-04-22)[2025-08-18]. https://link.springer.com/article/10.1186/s40537-025-01157-y.
[25]	SENANAYAKE J, KALUTARAGE H, AL-KADRI M O. Android Mobile Malware Detection Using Machine Learning: A Systematic Review[EB/OL].(2021-05-29)[2025-08-18]. https://www.mdpi.com/2079-9292/10/13/1606.
[26]	BILOT T, MADHOUN N E, AGHA K A, et al. A Survey on Malware Detection with Graph Representation Learning[J]. ACM Computing Surveys, 2024, 56(11): 1-36.
[27]	BABBAR H, RANI S, SAH D K, et al. Detection of Android Malware in the Internet of Things through the K-Nearest Neighbor Algorithm[EB/OL].(2023-07-20)[2025-08-18]. https://www.mdpi.com/1424-8220/23/16/7256.
[28]	DOLESI K, STEINBACH E, VELASQUEZ A, et al. A Machine Learning Approach to Ransomware Detection Using Opcode Features and K-Nearest Neighbors on Windows[EB/OL].(2024-10-11)[2025-08-18]. https://www.techrxiv.org/doi/full/10.36227/techrxiv.172926410.04244699.
[29]	GU Xiaoyao, YAN Jichang. Hierarchical K-Nearest Neighbors for Ransomware Detection Using Opcode Sequences[EB/OL].(2024-06-11)[2025-08-18]. https://www.techrxiv.org/doi/full/10.36227/techrxiv.171838524.46252988.
[30]	AL S A, ABUALHAJ M M, SHAMBOUR Q Y, et al. Enhancing Malware Detection Performance: Leveraging K-Nearest Neighbors with Firefly Optimization Algorithm[J]. Multimedia Tools and Applications, 2025, 84(12): 10071-10094.
[31]	ZHOU Shuncheng, LI Honghui, FU Xueliang, et al. A Novel Malware Detection Model in the Software Supply Chain Based on LSTM and SVMS[EB/OL].(2024-07-13)[2025-08-18]. https://www.mdpi.com/2076-3417/14/15/6678.
[32]	FARFOURA M E, MASHAL I, ALKHATIB A, et al. A Lightweight Machine Learning Methods for Malware Classification[EB/OL].(2024-10-10)[2025-08-18]. https://link.springer.com/article/10.1007/s10586-024-04755-2.
[33]	LI Junhao, HE Junjiang, LI Wenshan, et al. Syndroid: An Adaptive Enhanced Android Malware Classification Method Based on CTGAN-SVM[EB/OL].(2023-07-26)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S016740482300514X.
[34]	RIZVI S K J, ASLAM W, SHAHZAD M, et al. PROUD-MAL: Static Analysis-Based Progressive Framework for Deep Unsupervised Malware Classification of Windows Portable Executable[J]. Complex & Intelligent Systems, 2022, 8(1): 673-685.
[35]	CHAKRABORTY T, PIERAZZI F, SUBRAHMANIAN V S. EC2: Ensemble Clustering and Classification for Predicting Android Malware Families[J]. IEEE Transactions on Dependable and Secure Computing, 2017, 17(2): 262-277.
[36]	HAN Y S, SEO H B, YOON M K. Detecting Internet-of-Things Malware on Evidence Generation[EB/OL].(2024-08-06)[2025-08-18]. https://ieeexplore.ieee.org/abstract/document/10623818/.
[37]	LIU Zhen, WANG Ruoyu, JAPKOWICZ N, et al. Research on Unsupervised Feature Learning for Android Malware Detection Based on Restricted Boltzmann Machines[J]. Future Generation Computer Systems, 2021, 120: 91-108.
[38]	GUNDUZ H. Malware Detection Framework Based on Graph Variational Autoencoder Extracted Embeddings from API-Call Graphs[EB/OL].(2022-05-18)[2025-08-18]. https://peerj.com/articles/cs-988/.
[39]	DONG Shi, SHU Longhui, NIE Shan. Android Malware Detection Method Based on CNN and DNN Bybrid Mechanism[J]. IEEE Transactions on Industrial Informatics, 2024, 20(5): 7744-7753.
[40]	LIU Guiyun, ZHANG Jiezhao, ZHONG Xiaojing, et al. Hybrid Optimal Control for Malware Propagation in UAV-WSN System: A Stacking Ensemble Learning Control Algorithm[J]. IEEE Internet of Things Journal, 2024, 11(11): 36549-36568.
[41]	CHEN Jiaqi, YUAN Chong, LI Jiashuo, et al. ELAMD: An Ensemble Learning Framework for Adversarial Malware Defense[EB/OL].(2023-05-17)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S2214212623000923.
[42]	ZHU Huijuan, LI Yang, WANG Liangmin, et al. A Multi-Model Ensemble Learning Framework for Imbalanced Android Malware Detection[EB/OL].(2023-07-05)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0957417423014549.
[43]	VASHISHTHA L K, CHATTERJEE K, ROUT S S. An Ensemble Approach for Advance Malware Memory Analysis Using Image Classification Techniques[EB/OL].(2023-09-18)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S221421262300145X.
[44]	SINGH S, KRISHNAN D, VAZIRANI V, et al. Deep Hybrid Approach with Sequential Feature Extraction and Classification for Robust Malware Detection[EB/OL].(2024-09-26)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S1110866524001026.
[45]	THAKUR P, KANSAL V, RISHIWAL V. Hybrid Deep Learning Approach Based on LSTM and CNN for Malware Detection[J]. Wireless Personal Communications, 2024, 136(3): 1879-1901.
[46]	MORENO-TORRES J G, RAEDER T, ALAIZ-RODRIGUEZ R, et al. A Unifying View on Dataset Shift in Classification[J]. Pattern Recognition, 2012, 45(1): 521-530.
[47]	SHAHIDI S M, SHAKERI H, JALALI M. A Semantic Malware Detection Model Based on the GMDH Neural Networks[EB/OL].(2021-05-08)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0045790621001075.
[48]	ZHAO Wenxiang, WU Juntao, MENG Zhaoyi. Apppoet: Large Language Model Based Android Malware Detection via Multi-View Prompt Engineering[EB/OL].(2025-03-16)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0957417424024138.
[49]	OpenAI. Product and Research Updates of OpenAI[EB/OL].(2025-10-29)[2025-08-18]. Https://Openai.Com/.
[50]	ZHANG Yunchun, LI Haorui, ZHENG Yang, et al. Enhanced DNNs for Malware Classification with GAN-Based Adversarial Training[J]. Journal of Computer Virology and Hacking Techniques, 2021, 17(2): 153-163.
[51]	LI Qi, MI Jiaxin, LI Weishi, et al. CNN-Based Malware Variants Detection Method for Internet of Things[J]. IEEE Internet of Things Journal, 2021, 8(12): 16946-16962.
[52]	BENSAOUD A, KALITA J. CNN-LSTM and Transfer Learning Models for Malware Classification Based on Opcodes and API Calls[EB/OL].(2024-04-26)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0950705124001783.
[53]	KHAN S H, ALAHMADI T J, ULLAH W, et al. A New Deep Boosted CNN and Ensemble Learning Based Iot Malware Detection[EB/OL].(2023-10-08)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S016740482300295X.
[54]	HABIBI O, CHEMMAKHA M, LAZAAR M. Performance Evaluation of CNN and Pre-Trained Models for Malware Classification[J]. Arabian Journal for Science and Engineering, 2023, 48(8): 10355-10369.
[55]	JHA S, PRASHAR D, LONG H V, et al. Recurrent Neural Network for Detecting Malware[EB/OL].(2020-12-18)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404820303102.
[56]	CEN Mingcan, JIANG F, DOSS R. Ransoguard: A RNN-Based Framework Leveraging Pre-Attack Sensitive APIs for Early Ransomware Detection[EB/OL].(2025-03-16)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404824005996.
[57]	LUO Xi, FAN Honghui, YIN Lihua, et al. CAG-Malconv: A Byte-Level Malware Detection Method with CBAM and Attention-GRU[EB/OL].(2024-07-08)[2025-08-18]. https://ieeexplore.ieee.org/abstract/document/10589435/.
[58]	ABDULLAH M A, YU Y, ADU K, et al. HCL-Classifier: CNN and LSTM Based Hybrid Malware Classifier for Internet of Things (Iot)[J]. Future Generation Computer Systems, 2023, 142: 41-58.
[59]	OWOH N, ADEJOH J, HOSSEINZADEH S, et al. Malware Detection Based on API Call Sequence Analysis: A Gated Recurrent Unit-Generative Adversarial Network Model Approach[EB/OL].(2024-09-02)[2025-08-18]. https://www.mdpi.com/1999-5903/16/10/369.
[60]	WU Yinwei, LI Meijin, ZENG Qi, et al. Droidrl: Feature Selection for Android Malware Detection with Reinforcement Learning[EB/OL].(2023-05-16)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404823000366.
[61]	QIAN Liping, LIN Cong. Channel Features and API Frequency-Based Transformer Model for Malware Identification[EB/OL].(2023-12-07)[2025-08-18]. https://www.mdpi.com/1424-8220/24/2/580.
[62]	WANG Peng, LIN Tongcan, WU Di, et al. TTDAT: Two-Step Training Dual Attention Transformer for Malware Classification Based on API Call Sequences[EB/OL].(2023-11-11)[2025-08-18]. https://www.mdpi.com/2076-3417/14/1/92.
[63]	DEMIRKIRAN F, ÇAYIR A, UNAL U, et al. An Ensemble of Pre-Trained Transformer Models for Imbalanced Multiclass Malware Classification[EB/OL].(2022-10-16)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404822002401.
[64]	FENG Pengbin, GAI Le, YANG Li, et al. DawnGNN: Documentation Augmented Windows Malware Detection Using Graph Neural Network[EB/OL].(2024-05-13)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404824000890.
[65]	DENG Xiaoheng, WANG Zhe, PEI Xinjun, et al. Transmalde: An Effective Transformer Based Hierarchical Framework for Iot Malware Detection[J]. IEEE Transactions on Network Science and Engineering, 2023, 11(1): 140-151.
[66]	SHARMA O, SHARMA A, KALIA A. MIGAN: GAN for Facilitating Malware Image Synthesis with Improved Malware Classification on Novel Dataset[EB/OL].(2024-05-18)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0957417423031809.
[67]	KIM J Y, BU S J, CHO S B. Malware Detection Using Deep Transferred Generative Adversarial Networks[C]// Springer. International Conference on Neural Information Processing. Heidelberg: Springer, 2017: 556-564.
[68]	GAO Xiaowei, HU Changzhen, SHAN Chu, et al. Malicage: A Packed Malware Family Classification Framework Based on DNN and GAN[EB/OL].(2022-08-12)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S2214212622001296.
[69]	ZHANG Yunchun, LI Haorui, ZHENG Yang, et al. Enhanced DNNs for Malware Classification with GAN-Based Adversarial Training[J]. Journal of Computer Virology and Hacking Techniques, 2021, 17(2): 153-163.
[70]	KIM T G, KANG B J, RHO M, et al. A Multimodal Deep Learning Method for Android Malware Detection Using Various Features[J]. IEEE Transactions on Information Forensics and Security, 2018, 14(3): 773-788.
[71]	MENG Yi, LUKTARHAN N, YANG Xiaotong, et al. Gbadroid: An Android Malware Detection Method Based on Multi-View Feature Fusion[EB/OL].(2025-02-14)[2025-08-18]. https://link.springer.com/article/10.1007/s11227-025-06977-6.
[72]	CHEN Shaojie, LANG Bo, LIU Hongyu, et al. Android Malware Detection Method Based on Graph Attention Networks and Deep Fusion of Multimodal Features[EB/OL].(2024-05-06)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S095741742302119X.
[73]	SONG Yucai, CHEN Yang, LANG Bo, et al. Topic Model Based Android Malware Detection[C]// Springer. International Conference on Security, Privacy and Anonymity in Computation, Communication and Storage. Heidelberg: Springer, 2019: 384-396.
[74]	PENG Hao, YANG Jieshuai, ZHAO Dandan, et al. Malgne: Enhancing the Performance and Efficiency of Cfg-Based Malware Detector by Graph Node Embedding in Low Dimension Space[J]. IEEE Transactions on Information Forensics and Security, 2024, 19: 4881-4896.
[75]	WU Xiaowang, WANG Yan, FANG Yong, et al. Embedding Vector Generation Based on Function Call Graph for Effective Malware Detection and Classification[J]. Neural Computing and Applications, 2022, 34(11): 8643-8656.
[76]	PEKTAŞ A, ACARMAN T. Deep Learning for Effective Android Malware Detection Using API Call Graph Embeddings[J]. Soft Computing, 2020, 24(2): 1027-1043.
[77]	TANG Lihong, CHEN Xiao, WEN Sheng, et al. Demystifying the Evolution of Android Malware Variants[J]. IEEE Transactions on Dependable and Secure Computing, 2023, 21(4): 3324-3341.
[78]	AMER E, ZELINKA I. A Dynamic Windows Malware Detection and Prediction Method Based on Contextual Understanding of API Call Sequence[EB/OL].(2020-05-16)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404820300444.
[79]	GAO Han, CHENG Shaoyin, ZHANG Weiming. Gdroid: Android Malware Detection and Classification with Graph Convolutional Network[EB/OL].(2021-07-18)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404821000882.
[80]	YANG Jin, LIANG Huijia, REN Hang, et al. SAC: Collaborative Learning of Structure and Content Features for Android Malware Detection Framework[EB/OL].(2025-07-06)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0925231225007258.
[81]	GUO Wenjie, DU Wenbiao, YANG Xiuqi, et al. Malhapgnn: An Enhanced Call Graph-Based Malware Detection Framework Using Hierarchical Attention Pooling Graph Neural Network[EB/OL].(2025-01-10)[2025-08-18]. https://www.mdpi.com/1424-8220/25/2/374.
[82]	MENG Yi, LUKTARHAN N, YANG Xiaotong, et al. GBAdroid: An Android Malware Detection Method Based on Multi-View Feature Fusion[EB/OL].(2025-02-14)[2025-08-18]. https://link.springer.com/article/10.1007/s11227-025-06977-6.
[83]	GU Jintao, ZHU Hongliang, HAN Zewei, et al. GSEdroid: GNN-Based Android Malware Detection Framework Using Lightweight Semantic Embedding[EB/OL].(2024-05-16)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0167404824001081.
[84]	ZHANG Heyu, LI Binglong, YU Shilong, et al. Procgcn: Detecting Malicious Process in Memory Based on DGCNN[EB/OL].(2024-06-21)[2025-08-18]. https://peerj.com/articles/cs-2193/.
[85]	PENG Hao, YU Zehao, ZHAO Dandan, et al. Evading Control Flow Graph Based GNN Malware Detectors via Active Opcode Insertion Method with Maliciousness Preserving[EB/OL].(2025-03-17)[2025-08-18]. https://www.nature.com/articles/s41598-025-92023-7.
[86]	SUN Tiezhu, DAOUDI N, PIAN Weiguo, et al. Temporal-Incremental Learning for Android Malware Detection[J]. ACM Transactions on Software Engineering and Methodology, 2025, 34(4): 1-30.
[87]	RUNZE M, YIN Shangnan, FENG Xia, et al. A Lightweight Deep Learning-Based Android Malware Detection Framework[EB/OL].(2024-12-18)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S0957417424015008.
[88]	FERDOUS J, ISLAM R, MAHBOUBI A, et al. A Survey on ML Techniques for Multi-Platform Malware Detection: Securing PC, Mobile Devices, Iot, and Cloud Environments[EB/OL].(2025-02-13)[2025-08-18]. https://www.mdpi.com/1424-8220/25/4/1153.
[89]	GOPINATH M, SETHURAMAN S C. A Comprehensive Survey on Deep Learning Based Malware Detection Techniques[EB/OL].(2022-12-08)[2025-08-18]. https://www.sciencedirect.com/science/article/pii/S1574013722000636.

方法	典型方案	优点	缺点
KNN	文献[27,29]	原理简单且容易实现、对特征向量的适配性好	依赖样本分布均匀性、易受特征维度影响、样本量过大时距离计算耗时
SVM	文献[31]	在少样本场景鲁棒性高、处理非线性特征能力强	参数调优复杂、依赖数据的高质量标签、易受对抗攻击影响
聚类方法	文献[35]	无需标注数据、减少人工分析工作量、适合新型变种恶意检测	可解释性低、稳定性差、易出现簇边界模糊
降维方法	文献[37,38]	缓解“维度灾难”、减少计算与存储负担	合适的方法与有效的降维匹配困难、对特征的分布敏感
集成方法	文献[40,42]	对抗鲁棒性高、过拟合风险低、泛化能力强	方法复杂度高、依赖基学习器的质量、可解释性弱

方法	典型方案	优点	缺点
DNN	文献[48,47]	结构自适应、语义建模能力强	大规模标注数据与算力的依赖较强、可解释性差
CNN	文献[51,53]	特征学习自动化、对图像结构高敏感、泛化能力强	全局捕捉能力差、模型易过拟合、抗对抗样本弱
RNN	文献[56,58]	序列短程建模能力强、局部依赖捕捉能力强	对抗性弱、对长序列处理能力差
Transformer	文献[65,64]	全局依赖捕捉能力强、上下文理解能力好	对长序列敏感、模型训练量大
GAN	文献[66,68]	增强样本多样性、提升鲁棒性、辅助零日检测	训练不稳定、伪样本风险高、计算开销大
多模态	文献[70,71]	特征互补性强、鲁棒性高、检测性能好	模型结构复杂、模态对齐难度大

方法	典型方案	优点	缺点
图嵌入方法	文献[74,76]	可自动学习结构语义特征、支持低维表达与可视化、对下游分类任务有益	嵌入表达依赖图结构质量、特征维度设计敏感、训练过程存在过拟合风险
图聚类方法	文献[77]	无需标签、能学习变种演化、具备结构可解释性	计算量较大、容易受噪声影响、参数敏感
图神经网络	文献[82,83]	能学习语义与结构信息、捕捉局部与全局特征、适应复杂行为图谱	模型训练代价高、图扰动敏感、易遭对抗样本攻击

方法	优点	缺点	改进方向	应用场景
基于传统机器学习的方法	处理非线性关系能力强，能适应新型变种恶意行为，还能处理高维特征以及大规模数据，在保持精度的同时降低模型的训练量	高度依赖手工特征质量，泛化能力差，易受对抗样本影响，产生误报和漏报，且模型易老化，难以应对概念漂移	通过主动学习、数据增强等方式缓解高质量标注数据获取困难的问题，并持续优化特征提取流程以应对对抗性攻击和数据漂移问题	移动端、PC端的恶意软件识别
基于深度学习的方法	深度学习架构能自动提取高维、复杂特征，支持API序列、图像等多个模态输入，在大规模云环境下可高效处理海量样本，准确率高	训练需大量样本及高性能硬件支持，模型调参过程复杂，可解释性差，还存在过拟合风险	进行模型压缩及轻量化模型构建，引入可解释技术（如XAI、SHAP等）提升模型的可解释性	大规模样本检测、多模态分析等
基于图表示学习的方法	利用调用图、控制流图等结构捕捉程序之间的复杂依赖关系，擅长识别家族变种和行为异常，且具备一定可解释性，图结构便于追踪恶意逻辑路径	特征提取和图构建代价高，资源消耗大，且易受图结构对抗攻击影响，泛化能力有限	引入深度学习融合图结构，提升结构学习，同时利用剪枝、规约来优化图结构，还可进行对抗性训练增强鲁棒性	恶意家族溯源及变种恶意软件的识别