The research of ELM ensemble learning on multi-class resampling imbalanced data

The ELM has been proved to have good generalization performance and fast training speed in both theory and application. However, it tends to majority class and neglects minority class when dealing with imbalanced data. The Ensemble learning of data resampling can improve the ELM classification accuracy of a few classes. We propose a class resampling technique and advance an ELM ensemble learning method which can make use of the information of few class samples. Experimental results show that the proposed method is better than the single ELM learning model. Because resampling is one of the most core technologies of large data processing, the method provides the help for the establishment of the learning model of the imbalanced data.