Study on the Different Technique of Concept Drift and Novel Class Detection in Data Stream

Data streams mining has become interesting research topic and growing interest in knowledge discovery process. Because of the high speed and huge size of data and mining is processed with limited computing power and limited memory storage capabilities. Therefore our traditional classification technique are not directly applicable. Classification of data stream is more challenging task due to four major problems which is addresses by data stream mining: Infinite length, Concept-drift, Arrival of novel class and limited labeled data. In recent years great amount of work has been done to efficiently solve this problems. In this paper we discusses various technique which efficiently solve the problem of concept drift and novel class detection. Also we have present comparative analysis of this techniques.