Thank you! It's always best when you dig a little bit and find solution for yourself. That's the best way to learn.
Thanks for the Pack200 fix. Can you provide more info how the new project fails? Thanks
It seems like there is a bug in training workflow. Will put it on todo to fix for next release. Running training with only basic options should work
Colors correspond to neuron activation Red == 1 Greay == 0 Blue == -1 You can turn on display preferences to show values, and you can incpect indicvidual neurons in properties
Thats the right way, only when you drag test set dont clcik train, but test button in toolbar And if you click only test it wont destroy your allready trained network just make sure you've saved it before
Hi thanks for the note. It seems that here is some kind of misconfiguration on source forge host so many links ar ebroken, investigating... We're thinking of moving all tutorials to git hub so they could be updated by the community. In the meantime check out the code samples there is also wine example in Java code https://github.com/neuroph/neuroph/blob/master/neuroph-2.9/Samples/src/main/java/org/neuroph/samples/uci/WineClassificationSample.java Cheers
What is the output? Win / lose? Dont normalize as continious value since it doesent ake sense for this domain. Use one input neuron for each card, and binary input vector. Let us know if you get some interesting results
Did you entered the data or imported from csv file? You shoul dbe able to use th edat aset right away...