|
Simultaneous Feature and Sample Selection Using Ensemble Multi-objective Search Space Enhanced Modified Whale Optimization Algorithm |
|---|---|
| รหัสดีโอไอ | |
| Creator | M. Sathya, S. Manju Priya |
| Title | Simultaneous Feature and Sample Selection Using Ensemble Multi-objective Search Space Enhanced Modified Whale Optimization Algorithm |
| Contributor | - |
| Publisher | TuEngr Group |
| Publication Year | 2564 |
| Journal Title | International Transaction Journal of Engineering, Management, & Applied Sciences & Technologies |
| Journal Vol. | 12 |
| Journal No. | 6 |
| Page no. | 12A6C: 1-14 |
| Keyword | Microarray data analysis, F1-score, SMWOA, Feature Selection, Sample Selection, SFSS, Biological-inspired algorithm, Whale Optimization Algorithm (WOA), Levy Flight (LF), Pareto optimal problem, Cancer detection, Leukemia dataset, Support vector machine (SVM), Lymphoma, SFSS-EMSMWOA, Prostate, Lung cancer. |
| URL Website | http://TuEngr.com/Vol12_6.html |
| Website title | ITJEMAST V12(6) 2021 @ TuEngr.com |
| ISSN | 2228-9860 |
| Abstract | An effective method called Ensemble of Multi-objective Search space enhanced Modified Whale Optimization Algorithm (EMSMWOA) was used to solve the high dimensionality reduction problem in microarray data classification. In EMSMWOA, multiple SMWOA, evidential reasoning approach, and ensemble algorithm were utilized to choose the most important features, choose the optimal solution for feature selection from the Pareto-optimal set and generate optimal features respectively. The selected features were processed in different classifiers for microarray cancer detection. This work focuses on the computational complexity of classifiers also cut down by selecting features and samples of microarray data simultaneously using SMWOA. It also enhances the accuracy classification. After choosing the most relevant features and samples in every iteration, the Pareto optimal problem because of using the multiple objectives in SMWOAs is solved by applying the evidential reasoning approach. Finally, the best feature and sample subset are selected by applying the ensemble algorithm among multiple SMWOA that has various sizes of population and maximum iteration count. The results of SMWOA are ensemble depends on the mutual information in-between feature, sample, and class. The preferred features and samples are given as input to different classifiers for microarray cancer detection. The experiment proves the proposed SFSS-EMSMWOA method is improved on the basis of accuracy, precision, specificity, sensitivity, F1-score and also with average error than EMSMWOA in four different datasets for microarray data classification. |
