Optimization of Feature Weighting for Epitope Classification in B-Cell and SARS Using TVIWACRI-PSO-ELM

In the bio-molecular field, epitope classification is essential in vaccine development. A machine learning-based approach has been used for epitope classification using peptide data from b-cell, Severe Acute Respiratory Syndrome (SARS) to predict material for SARS-CoV-2 vaccine. This study aims to improve the performance of epitope classification by weighting peptide features and including input weight and biases of Classifier model using the Time Variant Inertia Weight, Acceleration Coefficients and Random Injection-Particle Swarm optimization (TVIWACRI-PSO) and the Extreme Learning Machine (ELM) classification algorithm. Experiment on b-cell, SARS dataset shows that feature weighting (FW) using TVIWACRI-PSO can improve accuracy performance by 8.4%.