Table 5.

CNN baseline model for MNIST dataset (with softmax output layer).

Layer	Details
Input layer	28 ×28 ×1 in [0.0, 1.0]
Convolution 1	1 ×5 ×5 kernels, ReLU,12 output maps of 28 ×28
Pooling 1	2 ×2 max-pooling, 12 output maps of 12 ×12
Convolution 2	12 ×5 ×5 kernels, ReLU,64 output maps of 12 ×12
Pooling 2	2 ×2 max-pooling, 64 output maps of 6 ×6
Flatten 1	Flatten, ReLU, 3,136 output maps of 1 ×1
Fully connected 1	Fully connected, ReLU, 100 output neurons
Fully connected 2	Fully connected, ReLU, 10 output neurons