how to choose layers in convet stacks for CNN operations ?

[Raady s]

We can see that Convolution Neural Network have combinations of multiple convets stacks( convolution layers and Pooling layers) for feature extraction  part and fully connected layers for classification. 

In the convolution layer how to decide the size of filters where convolution operations are performed ? Are there any papers or materials describing about the things happening inside the CNN for what reasons. 
I know mathematically how the CNN works and how to implement it. For example, there is a CNN network Lenet5 where the input size is 32x32, in convet1 there are 6 filters of size 28x28 and pooling layer of 14x14. How these sizes 32x32, 28x28 are choosen ? 

For some CNN structures have multiple convet stacks ( Lenet5 have two) and multiple Fully connected layers ( lenet 5 have two). How to decide how many convet stacks are required for feature extraction. 

How to decide the size of fully connected layer ? 

[Frédéric Bastien]

You need to do hyper-parameter search to find the best size of layers. Mostly, try many of them and pick the best!

Note, Pylearn2 don't have any developers anymore. There is other packages like Lasagne, Block and Keras build on top of Theano that are maintained.

Fred

--
You received this message because you are subscribed to the Google Groups "pylearn-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to pylearn-user...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.