Anomaly Detection Techniques - Study & Emergence of Novel Solutions for Network Anomaly Data

Data mining techniques make it possible to search large amounts of data for characteristic rules and patterns. If applied to network monitoring data recorded on a host or in a network, they can be used to detect intrusions, attacks and/or anomalies. In this paper, we present "Supervised & Unsupervised learning" a method to cascade K-means clustering and the Id3 decision tree learning methods to classifying anomalous and normal activities in a computer network. The K-means clustering method first partitions the training instances into two clusters using Euclidean distance similarity. On each cluster, representing a density region of normal or anomaly instances, we build an ID3 decision tree. The decision tree on each cluster refines the decision boundaries by learning the subgroups within the cluster. Our work studies the best algorithm by using classifyinganomalous and normal activities in a computer networks with supervised & unsupervised algorithms that have not been used before. We analyses the algorithm that have the best efficiency or the best learning and describes the proposed system of K-means&ID3 Decision Tree.