Research Article
Speech Deception Detection Based on EMD and Temporal Neural Network
Table 3
Results with signal decomposition reconstruction using different decomposition algorithms.
| | Methods | Original signal | With EMD | With LMD | With VMD | | Acc (%) | F1 (%) | Acc (%) | F1 (%) | Acc (%) | F1 (%) | Acc (%) | F1 (%) |
| | KNN | 77.3 | 77.0 | 78.8 | 78.5 | 78.2 | 78.5 | 76.3 | 76.0 | | SVM | 77.5 | 77.2 | 79.6 | 79.5 | 79.3 | 79.2 | 76.0 | 76.2 | | Decision tree | 70.0 | 70.7 | 70.8 | 71.0 | 71.0 | 70.5 | 69.4 | 69.0 | | Random forest | 78.6 | 79.0 | 79.6 | 79.3 | 79.0 | 78.8 | 78.1 | 78.3 | | Naïve Bayes | 68.8 | 69.1 | 70.3 | 70.1 | 70.1 | 70.2 | 66.9 | 66.6 | | AdaBoost | 72.8 | 73.1 | 72.9 | 73.1 | 73.4 | 73.2 | 72.0 | 71.3 | | Ensemble learning | 79.2 | 79.5 | 81.2 | 80.7 | 80.5 | 80.1 | 78.0 | 78.1 | | MLP | 72.1 | 72.3 | 76.6 | 76.2 | 75.2 | 75.5 | 70.5 | 70.6 | | RNN | 64.7 | 64.2 | 66.2 | 66.0 | 65.8 | 65.7 | 64.1 | 63.2 | | Autoencoder | 67.4 | 68.1 | 72.9 | 72.8 | 70.8 | 71.1 | 66.6 | 65.8 |
|
|
