Committee of Intelligent Machines — Unity in Diversity of #NeuralNetworks
Preetham V V
81

Hey Preetham, thanks for this post, it’s rich in intuitive & executable comprehension on the subject. I had a question regarding the exponential 2^H space in drop-out regularization, I apologize if it’s trivial. How capable & efficient is a network minimizing for error (with what is essentially a gradient search) in trimming down the problem space for selecting nodes?