detect-lstm-model
detect-lstm-model copied to clipboard
Published
20 hours ago •
zhangbincheng1997
zhangbincheng1997
detect malicious URL and Request (Bi-LSTM、Bi-LSTM + CNN、CNN + Bi-LSTM、CNN + Bi-LSTM + CNN)
detect-lstm-model
URL
| Model | Accuracy | Precision | Recall | F1 Score | Time per Epoch |
|---|---|---|---|---|---|
| Bi-LSTM | 99.51% | 99.54% | 99.81% | 99.67% | 1242ms |
| Bi-LSTM + CNN | 99.78% | 99.73% | 99.97% | 99.86% | 1185ms |
| CNN + Bi-LSTM | 99.60% | 99.75% | 99.72% | 99.73% | 326ms |
| CNN + Bi-LSTM + CNN | 99.54% | 99.75% | 99.64% | 99.69% | 318ms |
Request
| Model | Accuracy | Precision | Recall | F1 Score | Time per Epoch |
|---|---|---|---|---|---|
| Bi-LSTM | 82.30% | 80.81% | 99.93% | 89.23% | 4434ms |
| Bi-LSTM + CNN | 99.11% | 99.06% | 99.68% | 99.36% | 4406ms |
| CNN + Bi-LSTM | 99.10% | 98.92% | 99.79% | 99.35% | 1101ms |
| CNN + Bi-LSTM + CNN | 99.45% | 99.24% | 99.96% | 99.60% | 1124ms |
Architecture
Bi-LSTM
Bi-LSTM + CNN
CNN + Bi-LSTM
CNN + Bi-LSTM + CNN
General Architecture
Input Layer
character embedding √
word embedding ×
CNN Layer
Output Layer
Metadata
57
Stars
21
Forks
Watchers
Owner
zhangbincheng1997
Metadata
detect malicious URL and Request (Bi-LSTM、Bi-LSTM + CNN、CNN + Bi-LSTM、CNN + Bi-LSTM + CNN)