How can I implement SVM classification function on embedded platform after it has been trained in MATLAB.
1 view (last 30 days)
Show older comments
I have trained SVM with training data using the fitcsvm() function. SVMModel = fitcsvm(X,y,'KernelScale',.5,'Standardize',true,'OutlierFraction',0.05); with output: SVMModel =
ClassificationSVM
PredictorNames: {'x1' 'x2'}
ResponseName: 'Y'
ClassNames: 1
ScoreTransform: 'none'
NumObservations: 900
Alpha: [460x1 double]
Bias: -28.3591
KernelParameters: [1x1 struct]
Mu: [0.6169 0.1173]
Sigma: [0.1381 0.0132]
BoxConstraints: [900x1 double]
ConvergenceInfo: [1x1 struct]
IsSupportVector: [900x1 logical]
Solver: 'SMO'
I am satisfied with the results of predict() function, implemented as:
[~,score] = predict(SVMModel,[X1(:),X2(:)]);
Now I want to calculate the score of an individual sample on an embedded platform with limited resources. How may I program the SVM-Model as well as the classification function on a system with say, 1 MegaBytes maximum ram memory?
It is also not clear how may I know the parameters like 'Alpha' and 'Bias' as these will be needed when writing the classification function.
I have no training in machine learning and therefore I have only minimum understanding of the technicalities of SVM algorithm. Apologies if the question seems to naive.
0 Comments
Answers (0)
See Also
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!