News for hackers. Stuff that matters.

A comment on Tinker with a Neural Network in Your Browser. Parent comment

Re:Tinker with a Neural Network in Your Browser by tchow on Wednesday April 13, 2016 @09:30AM (#11487950)
How do you know to choose 6 hidden layers with 6 neurons each though? Why not 'x' hidden layers with 'j' neurons each? or some other random number?

Also how do you know to choose a ReLu instead of a Tanh activation?

[ Parent | on HN ]
Re:Tinker with a Neural Network in Your Browser by espadrine on Wednesday April 13, 2016 @09:36AM (#11487994)
ReLu gives good results for deep learning: http://jmlr.org/proceedings/papers/v15/glorot11a/glorot11a.p....

6 layers is the maximum that this demonstration allows, and they kept j small-ish to show that you don't need that many to have good results.

[ Parent | on HN ]