講演抄録/キーワード |
講演名 |
2016-11-04 10:35
Deep Learning for Ransomware Detection ○Aragorn Tseng・YunChun Chen・YiHsiang Kao・Tsungnan Lin(NTU) IA2016-46 |
抄録 |
(和) |
Ransomware is malware that installs covertly on a victim's computer or smartphone, executes a cryptovirology attack and demands a ransom payment to restore it. Ransomwares have been the most serious threat in 2016, and this situation continues to worsen. Because of high reward for Ransomwares, more and more Ransomware families appear, and it make us more difficultly to detect them. There are different signatures or behaviors among different families (i.e. Locky ,Cerber,Cryptowall .....)or versions (i.e. CryptXXX2.0 ,CryptXXX3.0) of Ransomwares, it will be wonderful if there has a way that can detect potential Ransomware threats.
In this paper, we use deep-learning method to detect Ransomwares. At first we introduce how we label the data with different behaviors and what features we choose. And we present our model for detecting various Ransomwares and prevent them from encrypting victim's data. Experimental evaluation demonstrates that our deep-learning model can detect latest Ransomwares in high-speed network timely. |
(英) |
Ransomware is malware that installs covertly on a victim's computer or smartphone, executes a cryptovirology attack and demands a ransom payment to restore it. Ransomwares have been the most serious threat in 2016, and this situation continues to worsen. Because of high reward for Ransomwares, more and more Ransomware families appear, and it make us more difficultly to detect them. There are different signatures or behaviors among different families (i.e. Locky ,Cerber,Cryptowall .....)or versions (i.e. CryptXXX2.0 ,CryptXXX3.0) of Ransomwares, it will be wonderful if there has a way that can detect potential Ransomware threats.
In this paper, we use deep-learning method to detect Ransomwares. At first we introduce how we label the data with different behaviors and what features we choose. And we present our model for detecting various Ransomwares and prevent them from encrypting victim's data. Experimental evaluation demonstrates that our deep-learning model can detect latest Ransomwares in high-speed network timely. |
キーワード |
(和) |
Ransomware / deep-learning / cyber-attack / machine-learning / / / / |
(英) |
Ransomware / deep-learning / cyber-attack / machine-learning / / / / |
文献情報 |
信学技報, vol. 116, no. 282, IA2016-46, pp. 87-92, 2016年11月. |
資料番号 |
IA2016-46 |
発行日 |
2016-10-27 (IA) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
IA2016-46 |
|