Where is the Auto-Encoder process in Caffe?

[hamlet11ito]

Simple question from a  Caffe biginner:.  Where is the Auto-Encoder process, one of the most essence of DL,  is implemented in "Caffe"?  In the training process?

 

[Jan C Peters]

I don't even understand your question: What do you mean by 'process'? Do you even know how caffe works? You define a network structure based on layers and blobs, and a solver configuration which defines the training parameters. The network configuration is very flexible, so you can create all kinds of different networks, also autoencoders. For an example using MNIST see caffe/examples/mnist/mnist_autoencoder.prototxt.

Jan

[Vimal Thilak]

Hey hamlet11ito,

As Jan pointed out, check out the mnist autoencoder example to get a feel for how AEs are implemented in caffe. 

[hamlet11ito]

thanks!!