Class-specific extreme learning machine based on overall distribution for addressing binary imbalance problem

Abstract Imbalanced classification problems have significantly unequal class proportions. Conventional extreme learning machine (ELM) gives the same importance to all the samples leading to the results which favor the majority class. To solve this intrinsic deficiency, modification of ELM have been developed like weighted ELM (WELM), WELM based on the overall distribution (ODW-ELM) etc.. We have also recently developed class-specific ELM (CS-ELM) in one of our recent works. It has been shown in this work that the derivation of the output weights, $\boldsymbol{\beta}$ is more efficient compared to class-specific cost regulation ELM (CCRELM) for handling the class imbalance problem. Motivated by CCR-ELM X. Luo et al. has proposed the classifier ODW-ELM which is also not efficient for the imbalance learning. In this work, a novel class-specific WELM based on overall distribution (OD-CSELM) and the kernelized version of OD-CSELM (OD-CSKELM) is proposed to address the binary class imbalance problem more effectively. OD-CSELM and OD-CSKELM are motivated by CS-ELM. In addition, the computational complexities of OD-CSELM and OD-CSKELM are significantly lower than WELM and kernelized WELM respectively. The proposed work is evaluated by using the benchmark real-world imbalanced datasets. The experimental results demonstrate that the proposed work gives good generalization performance in contrast with rest of the classifiers for class imbalance learning.