Github user avulanov commented on the pull request:
https://github.com/apache/spark/pull/1290#issuecomment-103140364
@loachli Do you set enough iterations? Adding hidden layers requires more
iterations. While I do not know which setting do you use exactly, could you try
the example from
https://github.com/apache/spark/pull/1290#issuecomment-100334613 ?
With regards to the back propagation code, I implemented chain rule of back
propagation. Good theoretical explanation can be found here:
http://www.slideshare.net/kuwajima/cnnbp. It is in general form, so no need to
change it if new layers are introduced. The code assumes that there is one
functional layer at the top that provides an error function. Error is then
back-propagated through all layers and delta is computed (the code you qouted).
Delta is later used to compute the gradient of weights.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
