Paper Abstract and Keywords |
Presentation |
2016-11-25 16:10
Performance of link-mining techniques to detect malicious websites Yasuhiro Takano, Daiki Ito, Tatsuya Nagai (Kobe Univ.), Masaki Kamizono (PwC Cyber Services), Masami Mohri (Gifu Univ.), Yoshiaki Shiraishi (Kobe Univ.), Yuji Hoshizawa (PwC Cyber Services), Masakatu Morii (Kobe Univ.) ICSS2016-44 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Conventional techniques to avoid malicious websites techniques by referring URL's keywords reported in black lists have been studied. Since attackers can modify the URL quite often, however, the conventional techniques are concerned that they are difficult to follow the frequent updates. Our previous contribution has shown that the malicious websites have a certain correlation among them. This paper evaluates, therefore, performance of supervised-inkmining techniques to detect the malicious websites by inputting the link structure captured from the actual websites. The experimental evaluation results shows that by determining the networks automatically the convolutional neural networks (CNN) algorithms achieves the accuracy = 87%, which outperform the support vector classification (SVC) techniques significantly. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
drive-by-download attack / linkmining / support vector classification (SVC) / convolutional neural networks (CNN) / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 116, no. 328, ICSS2016-44, pp. 31-35, Nov. 2016. |
Paper # |
ICSS2016-44 |
Date of Issue |
2016-11-18 (ICSS) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
ICSS2016-44 |
|