Á¤º¸°úÇÐȸ ÄÄÇ»ÆÃÀÇ ½ÇÁ¦ ³í¹®Áö (KIISE Transactions on Computing Practices)
Current Result Document : 574 / 575
ÇѱÛÁ¦¸ñ(Korean Title) |
ÇÕ¼º°ö½Å°æ¸ÁÀ» »ç¿ëÇÑ À̹ÌÁö ±â¹ÝÀÇ ¾Èµå·ÎÀÌµå ¾Ç¼º¼ÒÇÁÆ®¿þ¾î Æйи® ºÐ·ù |
¿µ¹®Á¦¸ñ(English Title) |
Image-based Android Malware Family Classification Using Convolutional Neural Network |
ÀúÀÚ(Author) |
°¹®¿µ
¹Ú¼ºÇö
¹ÚÁöÇö
Á¶¼ºÁ¦
¹Ú¹Î±Ô
Munyeong Kang
Seonghyun Park
Jihyeon Park
Seong-je Cho
Minkyu Park
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 27 NO. 04 PP. 0189 ~ 0197 (2021. 04) |
Çѱ۳»¿ë (Korean Abstract) |
¾Èµå·ÎÀÌµå ¾Ç¼º¼ÒÇÁÆ®¿þ¾î°¡ Áö¼ÓÀûÀ¸·Î Áõ°¡ÇÔ¿¡ µû¶ó, ±â°èÇнÀÀ» »ç¿ëÇÑ ¾Èµå·ÎÀÌµå ¾Ç¼º ¼ÒÇÁÆ®¿þ¾î ŽÁö ¹× ºÐ·ù ±â¹ýÀÌ ¸¹ÀÌ ¿¬±¸µÇ°í ÀÖ´Ù. ¾Ç¼º¼ÒÇÁÆ®¿þ¾î Æйи®(malware family) ºÐ·ù´Â, ¾Ç¼º¼ÒÇÁÆ®¿þ¾î »ùÇõéÀ» ¿¬°ü¼º ÀÖ´Â ±×·ìÀ¸·Î ºÐ·ùÇÏ´Â ±â¹ýÀ¸·Î ÄÄÇ»ÅÍ Æ÷·»½Ä ºÐ¼®, À§Çù Æò°¡, À§Çù ¿ÏÈ °èȹ¿¡ Áß¿äÇÑ ¿ªÇÒÀ» ÇÑ´Ù. º» ³í¹®¿¡¼´Â ½ÇÇàÆÄÀÏ ÁßÀÇ ÀϺθ¦ ȸ»öÁ¶ À̹ÌÁö(grayscale image) ·Î º¯È¯ÇÑ ÈÄ º¯È¯µÈ ¿µ»óµéÀ» ´ë»óÀ¸·Î µö·¯´× ±â¹ýÀ» Àû¿ëÇÏ¿© ¾Èµå·ÎÀÌµå ¾Ç¼º¼ÒÇÁÆ®¿þ¾î Æйи®¸¦ ºÐ·ùÇÏ´Â ¹æ¹ýÀ» Á¦¾ÈÇÑ´Ù. ´ëÇ¥ÀûÀÎ ¾Èµå·ÎÀÌµå ¾Ç¼º¼ÒÇÁÆ®¿þ¾î µ¥ÀÌÅÍ ¼Â(dataset)ÀÎ Drebin¿¡¼ Á¦°øµÇ´Â ¾Ç¼º¼ÒÇÁÆ®¿þ¾î ´ëÇ¥ Æйи®µéÀ» ´ë»óÀ¸·Î ÇÕ¼º°ö½Å°æ¸Á(Convolutional Neural Network, CNN) ¸ðµ¨À» Àû¿ëÇÏ¿© ¾Ç¼º¼ÒÇÁÆ®¿þ¾î¸¦ ºÐ·ùÇÑ´Ù. º» ½ÇÇèÀÇ ¿¬±¸ °á°ú¸¦ ±âÁ¸ ¿¬±¸ °á°ú¿Í ºñ±³ÇÏ¿©, µ¥ÀÌÅÍ °æ·®È¿Í ÀûÀýÇÑ µ¥ÀÌÅÍ Å©±âÀÇ ¼±Á¤, Á¤È®µµ¿¡ ÀÖ¾î º» ¿¬±¸°¡ ¾Ç¼º¼ÒÇÁÆ®¿þ¾î ºÐ·ù¿¡ ´õ È¿°úÀûÀÓÀ» º¸ÀδÙ.
|
¿µ¹®³»¿ë (English Abstract) |
As Android malware continues to increase, Android malware detection and classification techniques using machine learning are being studied intensively. Malware family classification is a technique for classifying malware samples into related malware families and plays an important role in computer forensic analysis, threat assessment, and threat mitigation planning. In this paper, we propose a method to classify Android malware families by converting only part of an executable file into a gray scale image and applying deep learning to the converted images. The malware samples are classified from the representative families of the dataset from the Drebin project by applying the Convolutional Neural Network (CNN) model. The experimental results show that the proposed method is more effective in classifying malware families in terms of processing overhead and classification accuracy.
|
Å°¿öµå(Keyword) |
¾Èµå·ÎÀÌµå ¾Ç¼º¼ÒÇÁÆ®¿þ¾î
Æйи® ºÐ·ù
ȸ»öÁ¶ À̹ÌÁö
ÇÕ¼º°ö ½Å°æ¸Á
µ¥ÀÌÅÍ ¼½¼Ç
android malware
family classification
grayscale image
CNN
data section
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|