It looks like you're new here. If you want to get involved, click one of these buttons!
Jumped straight in to take a look at the Deep learning operators by reviewing the examples and got this error.
the new 7.2.0 release of the Series Extension is adapted to RapidMiner 7.2.0, so you should not experience this security error with the Sliding Window Validation.
You can upgrade the extension from the Marketplace.
thank you for the report.
Would it be possible to share the process XML (.rmp content) with us? I see you are using Generate Data, so maybe there is no need to share any data.
Edit: Sorry, is this the unmodified "Regression using Deep Learning" tutorial process?
I am using one of the two embedded tutorials from within Rapidminer Studio. There is one for classification and one for regression.
That is certainly not the expected behaviour...
I saw your topic on the 7.2 update problem, is it completely resolved? (I could imagine this type of error, if somehow the installation is corrupted.)
Can you please share the rapidminer-studio.log file located in the .RapidMiner folder in your home folder after running into this issue? (Or the content of the Log Panel, but the former may contain more.)
I downloaded the 7.2 (x64) full installation, removed 7.1 and installed 7.2. There did not seem to be any problems. Apart from a a couple of processes that depended on some extentions that are not supported anymore, everything else seems to work. I also get the error with the Generalized Linear Model operator.
With Gradient Boosted Trees however, the first two tutorials work but the Third (Regression using GBT) fails with the same "Model Training Error (H2O)"
the log could help in figuring out if there is any error during Studio load and if there are more details to the error. You can also send me a PM.
The log outputs this when I run the process:-
Aug 9, 2016 4:47:45 PM SEVERE: Process failed: Error while training the H2O model: java.lang.RuntimeException: java.lang.ClassNotFoundException: water.Keyed$Icer
Aug 9, 2016 4:47:45 PM SEVERE: Here:
Aug 9, 2016 4:47:45 PM SEVERE: Process (Process)
Aug 9, 2016 4:47:45 PM SEVERE: subprocess 'Main Process'
Aug 9, 2016 4:47:45 PM SEVERE: +- Retrieve Titanic (Retrieve)
Aug 9, 2016 4:47:45 PM SEVERE: +- Set Role (Set Role)
Aug 9, 2016 4:47:45 PM SEVERE: +- Validation (Split Validation)
Aug 9, 2016 4:47:45 PM SEVERE: subprocess 'Training'
Aug 9, 2016 4:47:45 PM SEVERE: ==> | +- Deep Learning (Deep Learning)
Aug 9, 2016 4:47:45 PM SEVERE: subprocess 'Testing'
Aug 9, 2016 4:47:45 PM SEVERE: +- Apply Model (Apply Model)
Aug 9, 2016 4:47:45 PM SEVERE: +- Performance (Performance (Binominal Classification))
Rebooting and restarting Rapidminer allows the process to run. If I have a process that crashes the operator which I did then closing that process and opening a known working one still causes the operator to crash. I also had to manually kill the java process because I was running out of memory. If there is problem with the operator then simply closing the process and starting fresh might not be enough.
Thanks for the update.
If you are able to reproduce this again, please share the rapidminer-studio.log and h2o.log files located in the .RapidMiner folder (in your home folder).
(You cannot attach a file in personal message, but probably any file sharing web page can be used.)
This is what starts off the problem. I have a process that uses Logistic Regression (SVM) which works fine. Substituting H20 Logistic Regression gives this error:-
Process failed: Error while training the H2O model: java.security.AccessControlException: access denied ("java.lang.reflect.ReflectPermission" "suppressAccessChecks")
Once this happens, starting a known working process like the examples will fail until I restart Rapidminer and kill java manually.
Update: The new operators work but will always crash if used with sliding window validation.
Wow! That was fast. Thanks Peter!
I will take a look right now.
RapidMiner AI Hub
Automated Data Science
Training Classes & Certification
ML Algorithm Reference
Educational License Program