Synthetic minority oversampling technique for multiclass imbalance problems. (December 2017)

• Record Type: Journal Article
• Authors: Zhu, Tuanfei; Lin, Yaping; Liu, Yonghe
• Is Part Of: Pattern recognition. Volume 72(2017:Dec.)
• Page Start: 327
• Page End: 340
• Publication Date: 2017-12
• Subjects: Multiclass imbalance problems -- Synthetic minority oversampling -- Over generalization -- Neighbor directions; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2017.07.024
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: Synthetic oversampling technique for multiclass imbalance problems is proposed. Mechanism of avoiding over generalization is established. Regions of minority classes can aggressively enlarge. Effective minimization of class overlapping. Superior performance than the state of the art over various classifiers. Abstract: Multiclass imbalance data learning has attracted increasing interests from the research community. Unfortunately, existing oversampling solutions, when facing this more challenging problem as compared to two-class imbalance case, have shown their respective deficiencies such as causing serious over generalization or not actively improving the class imbalance in data space. We propose a k -nearest neighbors ( k -NN)-based synthetic minority oversampling algorithm, termed SMOM, to handle multiclass imbalance problems. Different from previous k -NN-based oversampling algorithms, where for any original minority instance the synthetic instances are randomly generated in the directions of its k -nearest neighbors, SMOM assigns a selection weight to each neighbor direction. The neighbor directions that can produce serious over generalization will be given small selection weights. This way, SMOM forms a mechanism of avoiding over generalization as the safer neighbor directions are more likely to be selected to yield the synthetic instances. Owing to this, SMOM can aggressively explore the regions of minority classes by configuring a high value for parameter k, but do not result in severe over generalization. Extensive experiments using 27 real-world data sets demonstrate the effectiveness of our algorithm.