| 236 | 0 | 28 |
| 下载次数 | 被引频次 | 阅读次数 |
恶意软件旨在破坏、禁用或控制计算机系统. Android恶意软件专门针对Android操作系统，以泄露机密信息和破坏系统为目的.文献显示相关领域已进行了多次尝试来检测Android恶意软件.然而，这些工作无法自动检测恶意软件，而且大多数都是基于签名的，无法检测恶意软件的新变种.本研究中，探索了不同的算法，以获得恶意软件预测的最佳算法，并获得有助于本研究有效预测恶意软件的最佳特征集.从本研究的分析中，已经看到，在预测恶意软件方面，集成方法比传统的机器学习算法要好.本研究使用LGBM创新算法将特征数量从215个减少到100个，精准率达到99.50%.此外，本研究使用只有55个特征的随机森林获得了99.17%的精准度.
Abstract：Malicious software aims to destroy, disable, or control computer systems. Android malware is specifically targeted at the Android operating system, with the aim of leaking confidential information and damaging the system. The literature shows that multiple attempts have been made in the relevant field to detect Android malware. However, these tasks cannot automatically detect malware, and most of them are signature-based, making it impossible to detect new variants of malware. In this study, different algorithms were explored to obtain the best algorithm for predicting malware and to obtain the best feature set that can help effectively predict malware. The analysis of this study shows that ensemble methods are better than traditional machine learning algorithms in predicting malware. The LGBM innovative algorithm is used in this study to reduce the number of features from 215 to 100, with an accuracy rate of 99.5%. In addition, an accuracy of 99.17% is achieved using a random forest with only 55 features.
[1] 199IT. Counterpoint：华为HarmonyOS系统在中国的市场份额已经达到8%[EB/OL].[2024-4-10]. http://www.199it.com/archives/1604885.html
[2] 199IT.卡巴斯基：2023年Q1移动设备攻击报告[EB/OL].[2024-4-12]. http://www.199it.com/archives/1623195.html
[3] KHARIWAL K,SINGH J,ARORA A. IPDroid:android malware detection using intents and permissions[C]//2020 Fourth World Conference on Smart Trends in Systems Security and Sustainablity(WorldS4).2020.
[4]王季，景齐，高建波，等. SEdroid：一种使用选择性集成学习的强大安卓恶意软件检测器[J]. IEEE无线通信与网络会议（WCNC）,2020:19-22.
[5] MAHINDRU A,SANGAL A L. MLDroid-framework for Android malware detection using machine learning techniques[J]. Neural Computing&Applications,2021(10):33.
[6]张雪芹，王逸璇，赵敏.基于深度学习的Android恶意软件动态检测[J].计算机工程与设计，2024,45(1):10-16.
[7] WANG W,GAO Z,ZHAO M,et al. DroidEnsemble:detecting android malicious applications with ensemble of string and structural static features[J]. IEEE Access,2018,6:31798-31807.
[8] FAHAD A,KHALED E. Android malware permission-based multi-class classification using extremely randomized trees[J]. IEEE Access,2018,6:76217-76227.
[9] CHANG W L,SUN H M,WU W. An android behavior-based malware detection method using machine learning[J]. IEEE International Conference on Signal Processing,Communications and Computing(ICSPCC),2016:1-4.
[10] YUAN Z L,LU Y Q,WANG Z G,et al. Droid-sec:deep learning in android malware detection[J].Computer Communication Review,2014,44(4):371-372.
[11] KUMARAN M,LI W J. Lightweight malware detection based on machine learning algorithms and the android manifest file[C]//IEEE Mit Undergraduate Research Technology Conference. IEEE,2016:1-3.
[12] SONALI K,PRAVIN K,VILAS T. Static analysis of android permissions and sms using machine learning algorithms[J]. International Journal of Computer Applications,2018,182(16):22-27.
基本信息:
DOI：
中图分类号:TP309;TP181
引用信息:
[1]贺军忠,安明明.基于集成机器学习算法的Android恶意软件创新预测方法研究[J].汕头大学学报(自然科学版),2024,39(04):65-73.
基金信息:
2023年甘肃省科学技术厅省级科技计划项目（23JRZK0524）; 2022年校级重点教学改革项目（JXGG2022001）