search

bobye / neuron

Scala library for neural networks
109 stars 32 forks source link

Stochastic Gradient Training with mini-batch mode #11

Open bobye opened 10 years ago

bobye commented 10 years ago

SGD is hard to use ...

Ref: http://yaroslavvb.blogspot.com/2014/03/stochastic-gradient-methods-2014.html

bobye commented 10 years ago

SGD with momentum works well.