Research Article
A Robust Continuous Authentication System Using Smartphone Sensors and Wasserstein Generative Adversarial Networks
Table 3
Accuracy (%) of handcrafted features and CNN-learned feature without any data augmentation.
| | Features | 10 s | 20 s | 30 s | 40 s | 50 s | 60 s |
| | Handcrafted features | 93.35 | 95.47 | 96.89 | 98.7 | 99.12 | 99.35 | | CNN-learned features | 98.12 | 99.25 | 99.36 | 99.45 | 99.58 | 99.52 | | Handcrafted features (DA) | 99.15 | 99.25 | 99.45 | 99.62 | 99.72 | 99.84 | | CNN-learned features (DA) | 99.35 | 99.46 | 99.56 | 99.72 | 99.88 | 99.98 |
|
|
