Research Article
Defending against Deep-Learning-Based Flow Correlation Attacks with Adversarial Examples
Table 2
The model hyperparameters of target model.
| | Layer | Details |
| | Convolution layer 1 | Kernel num: 2000 | | Kernel size: (2, 30) | | Stride: (2, 1) | | Activation: Relu |
| | Max pool 1 | Window size: (1, 5) | | Stride: (1, 1) |
| | Convolution layer 2 | Kernel num: 1000 | | Kernel size: (2; 10) | | Stride: (4, 1) | | Activation: Relu |
| | Max pool 2 | Window size: (1, 5) | | Stride: (1, 1) |
| | Fully connected 1 | Size: 3000, activation: Relu | | Fully connected 2 | Size: 800, activation: Relu | | Fully connected 3 | Size: 100, activation: Relu |
|
|
