I applied SVM classifier for my character recognition problem. It works nicely for two classes. But it can not be applied to multi class problem directly as in case of neural network. I have to classify Devnagari characters in 43 classes. Is the any approach or matlab code available?
As others have commented, there are many libraries implementing multi-class SVMs natively (personally, I recommend LibSVM, as in Olarik Surinta's answer).
However, if you do not want to use additional external software, you can easily construct a multi-class SVM starting from the binary SVM (actually, this is what most of the external libraries do). You can do this using different methods, the most popular being one-versus-all classification [1]. The idea is simply to train M binary SVMs, where M is the number of classes, and in the testing phase to output the class corresponding to the largest output between the M classifiers. You can implement this easily in Matlab. I think there is also some ready-to-use code on Matlab Central.
Once you become familiar with one-versus-all classification, there are other approaches you can additionally try, see [2].
[1] Rifkin, Ryan, and Aldebaro Klautau. "In defense of one-vs-all classification." The Journal of Machine Learning Research 5 (2004): 101-141.
[2] Hsu, Chih-Wei, and Chih-Jen Lin. "A comparison of methods for multiclass support vector machines." Neural Networks, IEEE Transactions on 13.2 (2002): 415-425.
http://www.loria.fr/~lauer/MSVMpack/MSVMpack.html
www.cmpe.boun.edu.tr/~ethem/files/papers/iwann99.pdf
www.researchgate.net%2Fpublication%2F2854340_LS-SVMlab_a_MATLABC_toolbox_for_Least_Squares_Support_Vector_Machines
Normally, I used LIBSVM for multi-class classification.
It supports MATLAB as well.
http://www.csie.ntu.edu.tw/~cjlin/libsvm/
See also "Cascading SVMs as a Tool for Medical Diagnosis using Multi-class Gene Expression Data", International Journal of Artificial Intelligence Tools, vol. 15, no. 3, pp. 335-352, http://www.worldscinet.com/ijait/15/1503/S0218213006002709.html, World Scientific, 2006.
As others have commented, there are many libraries implementing multi-class SVMs natively (personally, I recommend LibSVM, as in Olarik Surinta's answer).
However, if you do not want to use additional external software, you can easily construct a multi-class SVM starting from the binary SVM (actually, this is what most of the external libraries do). You can do this using different methods, the most popular being one-versus-all classification [1]. The idea is simply to train M binary SVMs, where M is the number of classes, and in the testing phase to output the class corresponding to the largest output between the M classifiers. You can implement this easily in Matlab. I think there is also some ready-to-use code on Matlab Central.
Once you become familiar with one-versus-all classification, there are other approaches you can additionally try, see [2].
[1] Rifkin, Ryan, and Aldebaro Klautau. "In defense of one-vs-all classification." The Journal of Machine Learning Research 5 (2004): 101-141.
[2] Hsu, Chih-Wei, and Chih-Jen Lin. "A comparison of methods for multiclass support vector machines." Neural Networks, IEEE Transactions on 13.2 (2002): 415-425.
Thanks all I am greatly enlightened that there is possibility any many approaches and research papers, code available
Thanks Mr. Simone Scardapane
I suggest you to search on Mathworks.com/MatlabCenteral
you can find many standard useful Matlab codes for your project, some might be added to your software and some not, but available for download.
Please explain the procedure for multiclass classification using LIBSVM libraray.
this link contains another tool for multiclass SVM
http://www.loria.fr/~lauer/MSVMpack/MSVMpack.html
Can LIBSVM use for multi-dimentional problems? Can any one guid me how to do that in Matlab.
Dear Isuru,
You may try fitrsvm function of matlab as well.
https://www.mathworks.com/help/stats/fitcsvm.html
matlap function
http://www.mathworks.com/matlabcentral/fileexchange/39352-multi-class-svm?focused=3776247&tab=function
Try these links:
https://www.mathworks.com/help/stats/classificationecoc.html;jsessionid=4226bbe8f0ba602eeed88b297f4d
https://www.mathworks.com/matlabcentral/fileexchange/39352-multi-class-svm
https://www.mathworks.com/matlabcentral/fileexchange/62061-multi-class-svm
- Badges
- Science topic