Initial RBM Implementation - Thesis

[shahinmv]

This is the initial implementation of RBM. Algorithm should be correct, tested each variable for rows and cols, also dot products just to be sure. Created new method for Neuronlayer class, for multiplying visible layer by weights of hidden layer. Initial implementation of it is commented out as I found out that the gemm function works perfectly for it.

[shahinmv]

I made the changes that you requested. About neuronlayer.cpp file, it still shows that I changed whole code. Only part changed is that I added dotProdVis(values, weights) method. All the matrices has been changed to const references. I added more comments on why I created new method inside neuron layer, also commented above in the pull request. Please let me know after you check it out.