Research Article
Convolution Neural Network-Based Sensitive Security Parameter Identification and Analysis
Table 3
Hyperparameters of CNN models.
| Hyperparameter | | |
| Filter size | {32, 64, 128, 256} | {4, 8, 16, 32} | Epoch | 20 | 15 | Output size (number of classes) | 5 | 4 | Batch size | 100 | 100 | Kernel size | | | Kernel initializer | he_normal | he_normal | Conv strides | | | Pooling strides | 2 | 2 | Output activation type | Softmax | Softmax | Optimizer | rmsprop | rmsprop | Loss function | categorical_crossentropy | categorical_crossentropy | Padding | Same | Same | Pooing sizes after each conv layer | | |
|
|