Subspace Detection on Concept Drifting Data Stream

Abstract

In recent years, data stream mining has become a hot spot in machine learning. Network data and sensor data are both data stream. However, there is concept drift problem in data stream so that traditional machine learning methods no longer work. Meanwhile, real-time learning is required in data stream and most of concept detection methods can’t support real-time demand. For solving this problem, this paper proposes a data stream learning framework which improves the classical Linear Discriminant Analysis (LDA) method based on a robust subspace learning method. It can not only detect concept drift in data stream quickly, but also classify data stream in real-time. The experimental results of sensor data and UCI repository validate the effectiveness of our method.