Deep-Learning-TensorFlow
Deep-Learning-TensorFlow copied to clipboard
The implement of CDK algorithm
Hi! I checked the CDK algorithm carefully, which is as below:
My question is: when using gaussin visible units, why the function 'compute_positive_association' uses 'hidden_probs'?And why it uses 'hidden_states' on binary visible units?
Thank you!
Hi! The reason is that I empirically found (on MNIST) that hidden_probs
works better with gaussian units while hidden_states
works better with binary units. But maybe this was only for MNIST..