ATS Drug Classification by Using Whale Optimization Based Descriptors
To improve the wrapper feature selection technique, swarm intelligence (SI) has been a preferred choice. The use of a binary whale optimization algorithm (BWOA) to handle the moleular descriptors selection problem in AMPHETAMINE-TYPE STIMULANTS (ATS) drug categorization has attracted this research. This work aims to improve the classifier's learning and prediction abilities in order to produce better classification results. BWOA are generated using S-shaped transfer functions, which are subsequently consolidated using a k-Nearest Neighbor (k-NN) classifier in the wrapper feature selection. Our goal is to see how different sigmoid transfer functions affect the significant feature selection and classification in BWOA. For performance assessment, several indicators and Wilcoxon's rank-sum test are used. The BWOA-S3 delivers performance improvements with the lowest fitness value, fast convergence, good classification accuracy, and a compact feature subset, according to experimental data. Three distinct classifiers also ratify the generalization of the best feature subset.