| Summary: | 碩士 === 國立中正大學 === 會計與資訊科技研究所 === 96 === Intrusion detection is a very important research issue in network security nowadays. Intrusion detection can be approached by data mining and machine learning techniques. In literature, advanced techniques by hybrid learning or ensemble methods have been considered, and they are superior to the models using single machine learning techniques. This thesis proposes a triangle area similarity measure combining the hybrid method, namely Triangle Area based Nearest Neighbors (TANN), in order to detect attacks more effectively. In TANN, we use k-means to obtain five cluster centers and transform data for k-NN classification by triangle area similarity measurement. By using KDDCup 99’ as the dataset and considering 10-fold cross validation, the experimental results show that TANN can effectively detect intrusion attacks and achieve higher detection and lower error rates than three baseline models based on support vector machines, k-NN, and the hybrid model combining k-means and k-NN.
|
|---|
