I was wondering if anyone was aware of some work done for Adaptive learning rate for Convolutional RBM training?
KyungHyun Cho published an algorithm for RBM (Enhanced Gradient and Adaptive Learning Rate for Training RBMs), but I haven't found any related work for Convolutional RBM.
Looking at the paper and its implementation, it looks like the candidate are chosen by maximizing the free energy over different parameters configuration. Could I do the same with a CRBM? I haven't either have found a formula for the free energy of the CRBM, but I don't think it would be that hard.