Research Article
LiDAR: A Light-Weight Deep Learning-Based Malware Classifier for Edge Devices
Table 5
The evaluation results on the workstation using the three classifiers.
| Dataset | Classifier | CPU (%) | RAM (MB) | Classification time (seconds) | F1-score |
| SMS | CNNc | 245.80% | 262.13 | 0.96 | 0.94 | CNNg | 168.90% | 256.27 | 0.94 | 0.89 | LiDAR | 37.90% | 83.30 | 0.09 | 0.90 | E-mail | CNNc | 4,451.80% | 1,986.491 | 14.04 | 0.99 | CNNg | 4,067.00% | 972.08 | 4.57 | 0.98 | LiDAR | 106.90% | 243.92 | 3.78 | 0.93 | Malware in 2019 | CNNc | 3,868.80% | 980.04 | 1.85 | 0.94 | CNNg | 3,693.00% | 835.86 | 1.22 | 0.90 | LiDAR | 97.00% | 146.55 | 0.43 | 0.94 | Malware in 2020 | CNNc | 3,862.50% | 944.33 | 1.57 | 0.99 | CNNg | 3,538.50% | 792.52 | 1.51 | 0.97 | LiDAR | 190.70% | 221.27 | 1.31 | 0.94 |
|
|