Research Article
A Hybrid Deep Neural Network for Electricity Theft Detection Using Intelligent Antenna-Based Smart Meters
Table 3
Mapping of limitations with solutions and validations.
| | Limitations | Proposed solution | Validation |
| | Inability to handle imbalanced data (L.1) | Synthetic minority oversampling technique (SMOTE) (S.1) | SMOTE has balanced the data, shown in Table 4 (V.1). | | Overfitting (L.2) | CNN-GRU (S.2) | The validation is given in Figures 7 and 8 (V.2) | | High false-positive rate (L.3) | PSO (S.3) | The validation is given in Figures 9 and 10 (V.3) |
|
|
