You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@madlib.apache.org by Frank McQuillan <fm...@pivotal.io> on 2018/04/09 18:57:39 UTC

Suggested usability improvements for MLP classification with minibatch

Hello MADlib folks,

I was testing MNIST dataset with minibatch preprocessor + MLP on master,
and could not get it to converge.   It turned out to be user error (me) and
not a problem with convergence at all, because I forgot to 1-hot encode the
dependent variable.

But I am wondering if other people might do the same thing that I did and
get confused, so I created 2 JIRAs that I hope will help out:

https://issues.apache.org/jira/browse/MADLIB-1226
https://issues.apache.org/jira/browse/MADLIB-1227

I'd suggest we do this for the upcoming release 1.14.

Please let me know if you don't think this is the right approach.

Frank