Tips

• Do you want to use only dropout? use a fully connected layers and set feed forward flag to ONLY_DROPOUT.
• DO you want to use only pooling? Create a convolutional layer and set the convolutional flag to NO_CONVOLUTION
• DO you want to use only 2D padding? Create a convolutional layer and set convolutional flag to NO_CONVOLUTION then the pooling dimensions should be 1,1
• If you want to use convolutional or transposed just use convolution or transposeed convolution in convolutional layer