Hidden layer sizes for RBM

Post Reply
Posts: 2
Joined: Sun Apr 04, 2021 8:00 pm

Hidden layer sizes for RBM

Post by jonahburke »


For this line in the RMB code,

dnn = DNN(D, [1000, 750, 500], K, UnsupervisedModel=RBM),

the second argument in the DNN function, denoting the hidden layer sizes, should be just [1000], as there will be only one hidden layer for RBM?
Site Admin
Posts: 57
Joined: Sat Jul 28, 2018 3:46 am

Re: Hidden layer sizes for RBM

Post by lazyprogrammer »

Thanks for your question.

The DNN class implements a "deep neural network" (hence DNN) which contains multiple RBMs (see the final argument).

It's not an RBM itself.

Check the implementation details in the lecture to see what each class is responsible for and what functionality they have.
Post Reply

Return to “Unsupervised Deep Learning in Python”