Multiclass classification in LabVIEW using SVM and one-vs-all method

[AustinNiki]

Dr Bo Fu is a Certified LabVIEW Developer and joined the Austin Consultants team in December 2013 as a Systems Engineer. Dr Fu joins us from the University of Nottingham where he completed his PhD and did a Postdoc in Electronic Engineering, developing a high-speed dynamic-sampling camera and a real-time control spatial light modulator.

Dr Fu is currently working on a very exciting project involving a combination of different AI techniques including Neural Networks, K-means and SVN to name a few. Dr Fu posts regularly on his  blog and has written the post below about using SVM to do multi-class classification.

Dr Fu explains:

“Today I came across a problem to use  SVM to do multi-class classification. The  toolkit downloaded from NI did not provide the ability to do multi-class classification with SVM but only for two classes (it’s quite a useful tool still). So I took use of the SVM VIs and made a multi-class version using one-vs-all method.

There is a good  tutorial on one-vs-all or one-vs-rest classification by Andrew Ng. We pick one class each iteration as Class A and make the rest classes as Class B. Only the test data that locate in Class A are allocated to the known class. Here is the code:

The original trained labelled data are classified as Class 0, 1, 2, … N. In the i-th iteration, only the data from Class i are re-classified to Class 1 and the rest data are re-classified to Class 0. When the test data locate in class 1 area, they are classified as Class i. Any unsorted data are left in Class -1.

When I test the performance of this one-vs-all classifier, the result seems fine.