Classification And Novel Class Detection In Concept-Drifting Data Streams Under Time Constraints

Abstract— Classification And Novel Class Detection In Concept-Drifting Data Streams Under Time Constraints. Data stream is continuous < Final Year Projects > and always change in nature. Data stream mining is the process of extracting knowledge form continuous data. Due to its dynamic changing nature it has some major challenges like infinite length, novel class detection and concept-drift. Data stream is infinite in length and we cannot store it for historical purpose. Concept drift means data changes rapidly over time and novel class define as new class appear in continuous data stream. Classification is the challenging task in data stream and existing data mining classifier cannot detect novel class until the classification models are not trained. Different classification and clustering based techniques are used to detect novel class in data stream. In this paper we have discuss the different techniques of novel class detection and its comparative analysis.