A Classification and Novel Class Detection Algorithm for Concept Drift Data Stream Based on the Cohesiveness and Separation Index of Mahalanobis Distance. (19th March 2020)

• Record Type: Journal Article
• Authors: Li, Xiangjun; Zhou, Yong; Jin, Ziyan; Yu, Peng; Zhou, Shun
• Other Names: Nistazakis Hector E. Academic Editor.
• Is Part Of: Journal of electrical and computer engineering. Volume 2020(2020)
• Publication Date: 2020-03-19
• Subjects: Computer engineering -- Periodicals; Electrical engineering -- Periodicals; 621.3905
• DOI: 10.1155/2020/4027423
• Languages: English
• ISSNs: 2090-0147

Abstract

Abstract : Data stream mining has become a research hotspot in data mining and has attracted the attention of many scholars. However, the traditional data stream mining technology still has some problems to be solved in dealing with concept drift and concept evolution. In order to alleviate the influence of concept drift and concept evolution on novel class detection and classification, this paper proposes a classification and novel class detection algorithm based on the cohesiveness and separation index of Mahalanobis distance. Experimental results show that the algorithm can effectively mitigate the impact of concept drift on classification and novel class detection.