Efficient Learning Approaches for Agents in Data Mining
E. Padmalatha, C. R. K. Reddy, Padmajarani · 2014
In this paper we present a technique for intrusion detection in distributed network. Here, we use the CVFDT for the identification of sort of intrusion and we use the CMAC neural network for identifying normal and abnormal data. Initially we train the dataset by calculating the radius of history concept which is necessary to identify the concept drift. In distribute network, there would be number of nodes which are represented as systems and the nodes are grouped separately by means of K-means algorithm. We choose the centroid of each group as agent and the agent would check the concept drift on corresponding nodes in the group. If a node has concept drift, the agent would transfer it to CVFDT to identify the sort of intrusion and if the node has no concept drift, the agent would transfer to CMAC to identify whether the data is normal or abnormal.