Re: Using ADAMs software

> Consider the attached Convolutional Neural Network example
>
> Could somebody advise whether one could code this using the above software
>
> If yes - I am having trouble relating the various layers shown to the
> 'actors' available in this software

ADAMS does not incorporate any deep learning frameworks directly. A
few years ago, I tried adding deeplearning4j (the same that
WekaDeepLeaning4j uses) as a module, but their API keep changing all
the time. which invalidated code all the time and made it impossible
to keep up with (a lot of wrapper code needed to be written in order
to make options accessible in the GUI).

The (currently) only avenue for adding deep learning support to ADAMS
is through the WekaDeepLearning4j package
(https://deeplearning.cms.waikato.ac.nz/).
I've never used that package, but from a quick glance at the document,
you will need to configure a Dl4jMlpClassifier classifier with the
layers that your network is made of:
https://deeplearning.cms.waikato.ac.nz/examples/train_cnn/

This setup, which is just another Weka classifier, can (in theory) be
used just like any other classifier within ADAMS then.

Please note: ADAMS uses its own package cache for Weka packages (split
by Weka version), i.e., you will need to install that package through
the Weka Package manager user interface that is available from ADAMS'
main menu.

Final note: for more ADAMS related questions ( like workflow, Weka
Investigator, etc), please use the relevant ADAMS mailing list.

Cheers, Peter
-- 
Peter Reutemann
Dept. of Computer Science
University of Waikato, Hamilton, NZ
Mobile +64 22 190 2375
https://www.cs.waikato.ac.nz/~fracpete/
http://www.data-mining.co.nz/
_______________________________________________
Wekalist mailing list -- [email protected]
Send posts to [email protected]
To unsubscribe send an email to [email protected]
To subscribe, unsubscribe, etc., visit https://list.waikato.ac.nz/postorius/lists/wekalist.list.waikato.ac.nz
List etiquette: http://www.cs.waikato.ac.nz/~ml/weka/mailinglist_etiquette.html