Multiclass classifier

[yrenard]

@the developers: is there any progress or agreement on how the multi-class classifiers will be implemented in the future as there was still no clear decision made in the above topic? How is the work of bpayan progressing, I thought he was working on an extension of the SVM algorithm back then? If I find some free days in september, maybe I'll look into it too.

Dear Dieter,

sorry for late reply here, haven't had time to read the whole discussion but still, about your question : unfortunately, Baptiste decided to leave INRIA. He is now working in a service company, nothing related to BCI or OpenViBE as far as I know. By the way, I'd like to post an old tutorial I wrote about adding new classifiers in OpenViBE.

Yann

[ddvlamin]

Dear Yann,

Thanks this should certainly help. I missed that on your blog.

Dear Rapten,

I'll try to explain it with a very concrete example.

Imagine the target class label is set as OVTK_StimulationId_Label_01 and the non target class label as OVTK_StimulationId_Label_02. Imagine you have two inputs in the voting classifier, connected to the outputs of two classifier processors. If the first classifier processor classifies the label as OVTK_StimulationId_Label_01 then the first input of the voting classifier will get one vote. If the second classifier processor does not output OVTK_StimulationId_Label_01 but another label it won't get a vote. If the second classifier does output the label OVTK_StimulationId_Label_01 then the second input of the voting classifier will also get a vote and there will be an ex-aequo. For this reason you can also set a number of repetitions. The previous steps are then simply repeated and the input with the most votes will win. The voting classifier then outputs a label depending on the setting Result class label base If this is OVTK_StimulationId_Label_01, then it will output OVTK_StimulationId_Label_01 if the first inputs wins, OVTK_StimulationId_Label_02 if the second one wins.

Concerning measuring accuracy, you could use the box http://openvibe.inria.fr/documentation/ ... asure.html or the confusion matrix http://openvibe.inria.fr/documentation/ ... atrix.html For a discussion on this topic: viewtopic.php?f=13&t=442&hilit=multi+class

I hope this helps.

Best regards,
Dieter Devlaminck