MC-MLDCNN: Multichannel Multilayer Dilated Convolutional Neural Networks for Web Attack Detection

Figure 7

The general structure of MC-MLDCNN: the model has three channels. The first channel has a kernel size of 2 with exponentially increasing dilation sizes as 1, 2, and 4. The second channel has a kernel size of 3 with the same dilation sizes. The third layer has a kernel size of 4 with the same dilation sizes as the other layers. Since the kernel sizes vary at each channel, the affected regions are different at each layer. For the sake of simplicity, the dilation sizes are increased incrementally rather than exponentially.