Table 1.

Description of state-of-the-art methods used for feature selection in the literature.

S.No.	Methods	Algorithm and Tools	Datasets	Results	Authors
1	HFS (FMs)	reliefF-SVM-RFE	Caltech-256	Accuracy: 96.14 % Run time:1715s	Zhou et al. [11]
2	HFS (FMs)	Chi2 and Anova, RF and Extra-tree RF, LR, KNN, and SVM.	HER2 image	Recall: 93.9 %, Specificity: 86.6 %, Accuracy: 90.3 %, Precision: 87.5 %, F1-score 90.6 %.	Aguilera et al. [12]
3	WMs	PCA and SGbSA (GbSA-PCA)	UCI (Iris and E. coli)	Run time: 81.38 s and 653.83 s for respective datasets	Hosseini et al. [49]
4	WMs	wPCA based MRbTA and SVM	Wisconsin diagnostic breast cancer, wine, Leukemia microarray	Accuracy: 92.27 %, 93.79 %, 90.29 % for respective datasets	Kim et al. [50]
5	HFS	IPCA, Gaussian and Super Gaussian	Liver Toxicity, Prostate cancer, Yeast metabolomic	Average of correctly identified non-zero loadings: 86.7 %, 87.7 %, 80.80 %	Yao et al. [50]
6	HFS	IPC, ICA, naıve Bayes and SVM	Wisconsin Breast Cancer, Wine, Crabs	Accuracy: 96.85 %, 98.90 %, 99.50 % for respective datasets	Reza et al. [51]
7	HFS (FMs)	ReliefF, Chi square, and Symmetrical techniques, GA, SVM	Microstructural images: Annealing twin, Brass/bronze, Ductile cast iron, gray cast iron, Malleable cast iron, Nickel-based superalloy, White cast iron	Overall Accuracy: 90.1 %	Khan et al. [19]
8	EMs	Tree-based genetic program (GP-FER)	DISFA, DISFA+, CK+, MUG	Average accuracy: 94.2 %	Ghazouani et al. [21]
9	WMs	HHBBO and SVM	RADARSAT 2 (NLCD 2006)	Overall accuracy: 96.01 %, Average accuracy: 93.37 %	Rostami et al. [3]
10	WMs	bPSO and bGWO, SVM, AlexNet, Vgg19, GoogleNet and ResNet.	COVID-19, normal, pneumonia X-ray images	Overall accuracy: 99.38 %	Canayaz et al. [30]
11	WMs	ACO, GA and TS, Fuzzy Rough set (ACTFRO) and GATFRO)	SRBCT, DLBCL, Breast, Leukemia, Swarm behaviour	Accuracy: 90.48 %, 97.41 %,83.33 %, 94.74 %, 86.68 % for respective datasets	Meenachi et al. [31]
12	WMs	GA and PSO with bagging, SVM and DT	NASA Metrics Data (MDP)	Accuracy: 84.4 %, 87.2 % respective methods	Wahono et al. [32]
13	WMs	GWO, Adaptive PSO and MLP, SVM, DT, KNN, NBC, RFC, LR.	UCI Machine Learning Repository	Accuracy: 96 % and 97 % respective methods	Le et al. [25]
14	WMs	Rough set and Scatter search, LR, DT and NN	Australian dataset, UCI Repository	Accuracy: 90.5 %, 83.4 % and 87.9 %	Wang et al. [26]
15	HFS (FMs)	Chi-Square, PCC, MI, NDS and GA	IDS dataset	Accuracy (99.48 %)	Dey et al. [33]
16	WMs	Modified DE, fuzzy approach and CNNs	University of California, Irvine (UCI)	Accuracy 83 %	Vivekanandan et al. [28]
17	WMs	Binary encoded SSA based on PCA-fastICA	11 datasets from UCI	Overall accuracy: 94.73 %	Shekhawat et al. [29]
18	WMs	MAF and CNN, HGSO algorithm, RF, SVM	ISIC 2017 and HAM10000 datasets	Overall accuracy: 92.22 % and 99.34 % respectively	Obayya et al. [39]
19	WMs	MobileNetv1, MobileNetv2, NASNetMobile, linear SVC, SVM	Real traffic scenes	Accuracy:87.4 %	Doğan et al. [14]
20	WMs	Binary BCS based on PCA-fastICA	11 datasets from UCI	Highest accuracy: 95.2 %	Pandey et al. [17]