[mlpack-svn] [MLPACK] #345: Sparse Autoencoder Module
MLPACK Trac
trac at coffeetalk-1.cc.gatech.edu
Tue Apr 15 00:59:15 EDT 2014
#345: Sparse Autoencoder Module
----------------------------+-----------------------------------------------
Reporter: siddharth.950 | Owner:
Type: enhancement | Status: new
Priority: major | Milestone:
Component: mlpack | Resolution:
Keywords: | Blocking:
Blocked By: |
----------------------------+-----------------------------------------------
Comment (by siddharth.950):
Thanks for checking out the code. I will incorporate those changes. The
sparsity parameter is already included in the implementation, it is called
'rho'. We can specify the amount of sparsity we need using 'rho'. In case
you don't want your model to be sparse, you can set 'beta' to zero. 'Beta'
acts the same way for the sparsity cost(KL divergence) as 'lambda' does
for regularization. I hope that is what you meant when you mentioned
sparsity.
--
Ticket URL: <http://trac.research.cc.gatech.edu/fastlab/ticket/345#comment:9>
MLPACK <www.fast-lab.org>
MLPACK is an intuitive, fast, and scalable C++ machine learning library developed at Georgia Tech.
More information about the mlpack-svn
mailing list