A Weighted Discriminative Extreme Learning Machine Design for Lung Cancer Detection by an Electronic Nose System

Abstract

This article presents a study on lung cancer detection based on electronic nose technology. The pattern recognition algorithm is extremely crucial for an electronic nose system, but the customary learning algorithms usually prefer the majority class for class imbalance learning due to the assumption of equal misclassification costs. To address this challenge, in this article, we propose a novel classification method named weighted discriminative extreme learning machine (WDELM) for lung cancer diagnosis. First, the WDELM assigns a different weight to each particular sample by using a flexible weighting strategy, which enables it to perform classification tasks with the unbalanced class distribution. Then, an alternating iterative algorithm is proposed to solve the convex objective function with a theoretical analysis presented. Finally, the effectiveness of the proposed method has been evaluated on lung cancer datasets and public datasets by comparing it with conventional methods. Experimental results have confirmed that the WDELM surpasses conventional methods and is competent for lung cancer diagnosis.