Á¤º¸Ã³¸®ÇÐȸ ³í¹®Áö ¼ÒÇÁÆ®¿þ¾î ¹× µ¥ÀÌÅÍ °øÇÐ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
³×Æ®¿öÅ© Æ®·¡ÇÈ µ¥ÀÌÅÍÀÇ Èñ¼Ò Ŭ·¡½º ºÐ·ù ¹®Á¦ ÇØ°áÀ» À§ÇÑ Àüó¸® ¿¬±¸ |
¿µ¹®Á¦¸ñ(English Title) |
A Pre-processing Study to Solve the Problem of Rare Class Classification of Network Traffic Data |
ÀúÀÚ(Author) |
·ù°æÁØ
½Åµ¿ÀÏ
½Åµ¿±Ô
¹ÚÁ¤Âù
±èÁø±¹
Ryu Kyung Joon
Shin DongIl
Shin DongKyoo
Park JeongChan
Kim JinGoog
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 09 NO. 12 PP. 0411 ~ 0418 (2020. 12) |
Çѱ۳»¿ë (Korean Abstract) |
Á¤º¸º¸¾ÈÀ» À§ÇÑ IDS(Intrusion Detection Systems)´Â Åë»óÀûÀ¸·Î ¼¸í±â¹Ý(signature based) ħÀÔŽÁö½Ã½ºÅÛ°ú ÀÌ»ó±â¹Ý(anomaly-based) ħÀÔ Å½Áö½Ã½ºÅÛÀ¸·Î ºÐ·ùÇÑ´Ù. ÀÌ Áß¿¡¼µµ ³×Æ®¿öÅ©¿¡¼ ¹ß»ýÇÏ´Â Æ®·¡ÇÈ µ¥ÀÌÅ͸¦ ±â°èÇнÀÀ¸·Î ºÐ¼®ÇÏ´Â ÀÌ»ó±â¹Ý IDS ¿¬±¸°¡ È°¹ßÇÏ°Ô ÁøÇàµÆ´Ù. º» ³í¹®¿¡¼´Â °ø°Ý À¯Çü ÇнÀ¿¡ »ç¿ëµÇ´Â µ¥ÀÌÅÍ¿¡ Á¸ÀçÇÏ´Â Èñ¼Ò Ŭ·¡½º ¹®Á¦·Î ÀÎÇÑ ¼º´É ÀúÇϸ¦ ÇØ°áÇϱâ À§ÇÑ Àüó¸® ¹æ¾È¿¡ ´ëÇØ ¿¬±¸Çß´Ù. Èñ¼Ò Ŭ·¡½º(Rare Class)¿Í ÁØ Èñ¼Ò Ŭ·¡½º(Semi Rare Class)¸¦ ±âÁØÀ¸·Î µ¥ÀÌÅ͸¦ À籸¼ºÇÏ¿© ±â°èÇнÀÀÇ ºÐ·ù ¼º´ÉÀÇ °³¼±¿¡ ´ëÇÏ¿© ½ÇÇèÇß´Ù. À籸¼ºµÈ 3Á¾ÀÇ µ¥ÀÌÅÍ ¼¼Æ®¿¡ ´ëÇÏ¿© Wrapper¿Í Filter ¹æ½ÄÀ» ¿¬À̾î Àû¿ëÇÏ´Â ÇÏÀ̺긮µå Ư¡ ¼±ÅÃÀ» ¼öÇàÇÑ ÀÌÈÄ¿¡ Quantile Scaler·Î Á¤±Ôȸ¦ ó¸®ÇÏ¿© Àü󸮸¦ ¿Ï·áÇÑ´Ù. ÁغñµÈ µ¥ÀÌÅÍ´Â DNN(Deep Neural Network) ¸ðµ¨·Î ÇнÀÇÑ ÈÄ TP(True Positive)¿Í FN(False Negative)¸¦ ±âÁØÀ¸·Î ºÐ·ù ¼º´ÉÀ» Æò°¡Çß´Ù. ÀÌ ¿¬±¸¸¦ ÅëÇØ 3Á¾·ùÀÇ µ¥ÀÌÅÍ ¼¼Æ®¿¡¼ ºÐ·ù ¼º´ÉÀÌ ¸ðµÎ °³¼±µÇ´Â °á°ú¸¦ ¾ò¾ú´Ù.
|
¿µ¹®³»¿ë (English Abstract) |
In the field of information security, IDS(Intrusion Detection System) is normally classified in two different categories: signature-based IDS and anomaly-based IDS. Many studies in anomaly-based IDS have been conducted that analyze network traffic data generated in cyberspace by machine learning algorithms. In this paper, we studied pre-processing methods to overcome performance degradation problems cashed by rare classes. We experimented classification performance of a Machine Learning algorithm by reconstructing data set based on rare classes and semi rare classes. After reconstructing data into three different sets, wrapper and filter feature selection methods are applied continuously. Each data set is regularized by a quantile scaler. Depp neural network model is used for learning and validation. The evaluation results are compared by true positive values and false negative values. We acquired improved classification performances on all of three data sets.
|
Å°¿öµå(Keyword) |
±â°èÇнÀ
Èñ¼Ò Ŭ·¡½º
ÁØ Èñ¼Ò Ŭ·¡½º
Àüó¸®
Ư¡ ¼±ÅÃ
Machine Learning
Rare Class
Semi Rare Class
Pre-processing
Feature Selection
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|