| DC 欄位 |
值 |
語言 |
| DC.contributor | 資訊管理學系 | zh_TW |
| DC.creator | 葉季霏 | zh_TW |
| DC.creator | Chi-Fei Yeh | en_US |
| dc.date.accessioned | 2021-7-27T07:39:07Z | |
| dc.date.available | 2021-7-27T07:39:07Z | |
| dc.date.issued | 2021 | |
| dc.identifier.uri | http://ir.lib.ncu.edu.tw:88/thesis/view_etd.asp?URN=108423042 | |
| dc.contributor.department | 資訊管理學系 | zh_TW |
| DC.description | 國立中央大學 | zh_TW |
| DC.description | National Central University | en_US |
| dc.description.abstract | 人手一機是現今生活的寫照，不僅如此，行動裝置通常儲存大量使用者相關資料， 針對這些資料，當然存有專屬於個人所擁有之隱私資料，除此之外，便利的支付方式， 使得行動裝置也同時存有信用卡、金融卡等資料，對於大家每天使用的行動裝置安全， 也在近年來顯得更為重要。本論文主要針對時下市占率最高之行動裝置作業系統 Android 進行安全分析，透過靜態分析，並將 APK 檔案中的原始資料，轉換為圖像資 訊，使得可以快速分析出惡意軟體，圖像將結合該 APK 所使用到的權限與操作碼，後續組成 RGB 彩色圖像，下游分類器使用深度學習中，對於圖 像分類表現優異之卷積神經網路(Convolutional Neural Network)，使用 Autoencoder 加上 Efficient Net 架構，實驗於 CIC2020 與 Android Malware Dataset 兩種資料集，並進行惡 意軟體四類別以及惡意軟體家族分類，準確率雙雙達到 97%、F1-Score 達到 96%。 | zh_TW |
| dc.description.abstract | A hand-held device is a portrayal of today’s life. Mobile devices usually store a large amount of user-related data. For these data, there are of course private data owned by individuals. In addition, convenient Mobile payment methods make the device also stores information such as credit cards and financial cards, which has become more important in recent years for the safety of mobile devices that everyone uses every day. This research mainly conducts security analysis for Android, the mobile device operating system with the highest market share nowadays. Through static analysis, the original data in the APK file is converted into image information, so that malicious software can be quickly analyzed. The image will be combined with the permission and operation code used by the APK, and then the RGB color map is formed. The downstream classifier uses the ＂Convolutional Neural Network＂ that performs well in image classification in deep learning. Using Autoencoder and Efficient Net architecture, experimenting with two data sets, CIC2020 and Android Malware Dataset, and categorizing the four categories of malware and malware families, the accuracy rate reached 97%, and the F1-Score reached 96%. | en_US |
| DC.subject | Android | zh_TW |
| DC.subject | 惡意軟體偵測 | zh_TW |
| DC.subject | 靜態分析 | zh_TW |
| DC.subject | 惡意軟體圖像化 | zh_TW |
| DC.subject | 深度學習 | zh_TW |
| DC.subject | Android | en_US |
| DC.subject | Malware detection | en_US |
| DC.subject | Static analysis | en_US |
| DC.subject | Malware image method | en_US |
| DC.subject | Deep learning | en_US |
| DC.title | 一種多特徵 RGB 圖像表示法結合深度學習之 Android 惡意軟體偵測研究 | zh_TW |
| dc.language.iso | zh-TW | zh-TW |
| DC.title | A multi-feature RGB image representation combined with deep learning Android Malware detection research | en_US |
| DC.type | 博碩士論文 | zh_TW |
| DC.type | thesis | en_US |
| DC.publisher | National Central University | en_US |