Hi there,
I built an motor imagery classifier program and I would like to use AAR, Adaptative AutoRegressive method to the classifier phase. However i have no idea if i need to create a new classifier algorithm box or add the method somewhere in the algorithm boxes already existent.
For example I see the box algorithm "Classifier Trainer" has an attribute property called "Classifier Trainer to use"
and its value is set to LDA - linear discriminant analysis.
What should i do?
I wonder if that would be be a new and good feature to contribute to openvibe library
AAR classifier method
Re: AAR classifier method
Iuri,
the documentation lacks some information about how to create new classifiers. The best you could do for now is to look at the LDA source code and take some ideas from it. For this to be done, you should have a look to those files :
openvibe-plugins/classification-gpl/trunc/src/algorithms/classification/ovpCAlgorithmClassifierLDA.h
openvibe-plugins/classification-gpl/trunc/src/algorithms/classification/ovpCAlgorithmClassifierLDA.cpp
openvibe-plugins/classification-gpl/trunc/src/ovp_main.cpp
The LDA classifier is quite simple so you should be able to adapt the new files based on this example.
Keep attention to what is done in the ovp_main.cpp file... The classification algorithm is registered as an enumeration entry, making it available as a setting value to the classifier processor / trainer boxes.
Tell me if you need more input on this and feel free to join us on the IRC channel if you need live help
Yann
the documentation lacks some information about how to create new classifiers. The best you could do for now is to look at the LDA source code and take some ideas from it. For this to be done, you should have a look to those files :
openvibe-plugins/classification-gpl/trunc/src/algorithms/classification/ovpCAlgorithmClassifierLDA.h
openvibe-plugins/classification-gpl/trunc/src/algorithms/classification/ovpCAlgorithmClassifierLDA.cpp
openvibe-plugins/classification-gpl/trunc/src/ovp_main.cpp
The LDA classifier is quite simple so you should be able to adapt the new files based on this example.
Keep attention to what is done in the ovp_main.cpp file... The classification algorithm is registered as an enumeration entry, making it available as a setting value to the classifier processor / trainer boxes.
Tell me if you need more input on this and feel free to join us on the IRC channel if you need live help
Yann