Regularised extreme learning machine with misclassification cost and rejection cost for gene expression data...

Area of Science:

• Bioinformatics
• Machine Learning
• Computational Biology

Background:

• Traditional classification algorithms prioritize accuracy but fail to minimize misclassification costs.
• Accurate cost-sensitive classification is crucial in bioinformatics applications like tumor diagnosis.

Purpose of the Study:

• To propose a novel cost-sensitive regularised extreme learning machine (CS-RELM) algorithm.
• To minimize average misclassification costs by improving accuracy for high-cost samples.
• To integrate 'rejection cost' for further reduction of misclassification expenses.

Main Methods:

• Developed a CS-RELM algorithm incorporating probability estimation and misclassification costs.
• Integrated a 'rejection cost' mechanism into the CS-RELM algorithm.
• Evaluated CS-RELM on Colon Tumour and SRBCT datasets, comparing it with ELM, cost-sensitive ELM, regularised ELM, and cost-sensitive SVM.

Main Results:

• CS-RELM demonstrated superior performance in reducing average misclassification costs compared to other cost-sensitive algorithms.
• The inclusion of rejection cost further enhanced the algorithm's ability to minimize misclassification expenses.
• CS-RELM provided more credible classification decisions than benchmark algorithms.

Conclusions:

• The proposed CS-RELM algorithm effectively minimizes average misclassification costs in bioinformatics.
• CS-RELM offers a more reliable approach to classification, especially for datasets with varying misclassification costs.
• This method advances cost-sensitive learning in biological data analysis.