An adaptive resampling algorithm based on CFSFDP

摘要

This paper presents a novel adaptive resampling algorithm based on the clustering by fast search and find of density peaks (CFSFDP) algorithm and the synthetic minority oversampling technique (SMOTE), named DP-SMOTE. The essential idea of the proposed method is to use the improved CFSFDP algorithm to find the subclasses and removing noisy data automatically, and then to generate the minority samples within minority subclasses to prevent the synthetic samples falling inside the majority regions, with giving the boundary data higher oversampling weights. Experiments on UCI datasets show that the proposed DP-SMOTE algorithm is more efficient and adaptive than other oversampling algorithms.