Data Processing in Static Classifier Ensemble for Positive and Unlabeled Data Stream Classification Based Invasion Detection
Yan Li · Advanced materials research · 2014
In the research of invasion detection, Positive and Unlabeled Learning algorithms canreduce the amount of work for labeling training samples. The present data stream classificationalgorithms aim at totally labeled data stream. From the perspective of data stream, a novel invasiondetection algorithm which is based on positive and unlabeled data stream classification using staticclassifier ensemble is proposed in this chapter. The experimental results on different datasetsdemonstrate that the proposed invasion detection algorithm can achieve good detectionperformance with reduced labeled training samples.