Joint Sample Position Based Noise Filtering and Mean Shift Clustering for Imbalanced Classification Learning

The problem of imbalanced data classification learning has received much attention. Conventional algorithms are susceptible to skew favor majority samples and ignore minority samples. Majority weighted oversampling technique (MWMOTE) is an effective approach solve this problem, however, it may suffer from the shortcomings inadequate noise filtering synthesizing same as original data. To end, we propose improved MWMOTE method named joint sample position based mean shift clustering (SPMSC) these problems. Firstly, in order effectively eliminate effect noisy samples, SPMSC uses a new mechanism determine whether or not on its distribution relative sample. Note that generate duplicate then employ algorithm cluster reduce synthetic replicate Finally, cleaning performed processed further class overlap. Experiments extensive benchmark datasets demonstrate effectiveness compared with other sampling methods.