A Feature Weighted Ensemble Classifier on Stream Data

Wenhua Xu,Zheng Qin,Lei Ji,Yang Chang
DOI: https://doi.org/10.1109/CISE.2009.5364407
2009-01-01
Abstract:Stream data classification is a research topic of growing interest. Traditional approaches treat all attributes (i.e. features) of a stream data object fairly during the process of classification. Yet, in a real streaming environment, not all of the features are equally important to the classification result. Therefore, the classification accuracy can be improved by highlighting representative features and dimming irrelevant features. We apply a feature weighted ensemble classification model to solve this problem. The model is built using a modified K-means clustering technique and classification is performed with K-nearest neighbor algorithm. Experiments show that the method can improve the accuracy of classification, especially when there are noise features in high-dimensional stream data.
What problem does this paper attempt to address?