I have seen several different architectures for convolutional neural network (CNN). I am confused which one is the standard and how do I decide what to use. I am not confused by the number of layers being used or the number of parameters involved; I am confused by the COMPONENTS of the network.

Let assume:

    CL = convolution layer SL = subsampling layer(pooling) CM = convolution map NN = neural network Softmax = softmax classifier (similar to linear classifier)

Architecture 1


CL --> SL --> CL --> SL --> CM --> Softmax

Architecture 2 (Do we really need NN at the end again?)


CL --> SL --> CL --> SL --> NN --> Softmax

Architecture 3

My idea CL--> SL --> CL --> SL --> Softmax


After in-depth research, there are not a 100% guideline saying that you need to build,e.g., 5 layers of convolution + 2 layers of pooling + 3 layers of fully connected network, in order to achieve good performance.

As papers and competitions result showed (such as ImagNet compeition shows), the deeper the network is (>10 layers), the better the classification is assuming you have enough training data.

For general idea of what components are necessary in a network, you can, for example, read: 2009 What is the best multi-stage architecture for object recognition ny Karrett K., et. al. .


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.