images have the “stationarity” property, which implies that features that are useful in one region are also likely to be useful for other regions.

 

 

 

Convolutional networks may include local or global pooling layers^{[clarification needed]}, which combine the outputs of neuron clusters at one layer into a single neuron in the next layer.^[9][10] For example, max pooling uses the maximum value from each of a cluster of neurons at the prior layer.^[11] Another example is average pooling, which uses the average value from each of a cluster of neurons at the prior layer.

 

 

 

throwing away too much information

 

http://deeplearning.net/tutorial/lenet.html

 

 

 

http://ufldl.stanford.edu/tutorial/supervised/Pooling/