Table 8.
Comparison of the proposed method with other related works for the Bonn dataset.
| Work | Preprocessing | Feature Extraction | Feature Selection | Classifiers | Accuracy |
|---|---|---|---|---|---|
| [21] | TQWT | CCEnt | PCA | LS-SVM | 97.02% |
| [22] | TQWT | Hybrid Features | Firefly | RF | 97% |
| [23] | TQWT | AVP, STD | No | K-NN | 98.80% |
| [24] | TQWT | Statistic Features | No | K-NN | 100% |
| [25] | TQWT | KNN Entropy | Wrapper | SVM | 100% |
| [26] | TQWT | CTM, 2D-RPS plots | N/A | NA | N/A |
| [27] | TQWT | MvFE | No | LS-SVM | 84.67% |
| [28] | EMD–TQWT | IP | Different Methods | LS-SVM | 99% |
| [29] | TQWT | SC, SS, SF, SSl | No | bootstrap | 100% |
| [30] | TQWT | Correntropies | N/A | RF | 92.78% |
| [31] | TQWT | KnnEnt, CCorrEnt, FzEnt | No | LS-SVM | 95% |
| [32] | TQWT | Centered correntropy | No | RF | 98.30% |
| [33] | TQWTRF | FDs, AppEnt | No | SVMRF | 100% |
| [34] | TQWT | Mixture Correntropy | Various Methods | LS-SVM | 90.10% |
| [35] | IEVDHM–HT | Various Features | Student’s t-test | LS-SVM | 100% |
| [36] | FAWT | CVDistEnt, logarithmic energy | N/A | FKNN | 100% |
| [37] Multi-Classes = 99.46% |
VMD, HT | BLIMFs | No | EMRVFLN | Two-Classes = 100% Multi-Classes = 99.46% |
| [38] Multi-Classes = 96.50% |
Filtering | LSP | NCA | SVM | Two-Classes = 99.10% |
| [39] Multi-Classes = 99.70% |
Filtering, DWT | Different Features | N/A | SVM | Two-Classes = 99.50% Multi-Classes = 99.70% |
| [40] | DWT | Linear and Non-Linear Features | No | SVM | 99.50% |
| [41] | DWT | Statistic Features, Entropy, RWE | WOA | SVM | 99.80% |
| [42] | SSA | 1D-LBP | No | SVM | N/A |
| [43] | DWT | Entropy Features | ANOVA-FSFS | SVM | 99.50% |
| [44] Multi-Classes = 99.07% |
WPT | FDE | Kruskal Wallis | KNN | Two-Classes = 99.69% Multi-Classes = 99.07% |
| [45] | MODWPT | Statistic Parameters | Different Methods | LS-SVM | 99.60% |
| [46] | FSST | GLCM | N/A | KNN | 99.59% |
| [47] | ECT | Graph Theory, FD | No | RF | 98.50% |
| [48] | MRBF–MPSO | PSD | PCA | SVM | 98.73% |
| [49] | Z-Score Normalization | 1D-CNN | No | Softmax | 86.67% |
| [50] | DWT | PSR | SVCM | LS-SVM | 98.55% |
| [51] | EMD | Spectral and Temporal Features | No | SVM | N/A |
| [52] | ATFFWT | FD | Different Methods | LS-SVM | Two-Classes = 100% |
| [53] Multi-Classes = 100% |
TWD | Statistical Features | No | KNN | Multi-Classes = 100% 99.33% |
| [54] | DWT | Statistical Features | N/A | SVM | Two-Classes = 97.97% |
| [55] Multi-Classes = 98% |
IMFs | AmE | DESA | RF | Multi-Classes = 98% Two-Classes = 99.41% |
| [56] | DoG | LBP and Histogram Features | No | SVM | Multi-Classes = 98.80% 99.12% |
| [57] | GST | SVD Feature | No | RF | 97.78% |
| [58] | DCT | HE and ARMA Model | No | LSTM | 96% |
| [59] | DWT | Feature Extraction | No | N/A | 99.26% |
| [60] | -- | ApEn and RQA | No | N/A | 95% |
| [61] | WT | Approximate Entropy, LLE, Correlation Dimension | FRBS | N/A | 99% |
| [62] | Clustering, Covariance Matrix | Statistical Features | Non-Parametric Tests | AB-LS-SVM | Two-Classes = 99.64% |
| Proposed Method | TQWT | Statistical + Frequency + Fractal and Entropy Features | Proposed Convolutional RNN (CNN–RNN) | Multi-Classes = 99.71% | |