U09CO202 U09CO207
U09CO206 U09CO240
Outline
● Project Preliminary : A quick recap ● Running the SVM classifier - Weka ● Improvising the baseline model ○ Principal Component Analysis ○ Feature Subset Selection ● Comparison of different models ● Building a local database ● Next Steps...
Outline
● Project Preliminary : A quick recap ● Running the SVM classifier - Weka ● Improvising the baseline model ○ Principal Component Analysis ○ Feature Subset Selection ● Comparison of different models ● Building a local database ● Next Steps...
So far ...
● Used the Berlin speech emotion database Emo-DB: http://www.expressive-speech.net/emodb/ ● Feature extraction : openSMILE ● Trained multi-class SMO classifer in Weka ○ Accuracy : 82.4%
Workflow
Outline
● Project Preliminary : A quick recap ● Running the SVM classifier - Weka ● Improvising the baseline model ○ Principal Component Analysis ○ Feature Subset Selection ● Comparison of different models ● Building a local database ● Next Steps...
Weka SMO parameters
We used the foll. configuration in the Classify window of the Weka Explorer : weka . classifiers . functions . SMO −C 1 . 0 −L 0 . 0 0 1 −P 1 . 0 E−12 −N 0 −V −10 −W 1 −K ” weka .classifiers . functions . supportVector . PolyKernel −C 250007 −E 1 . 0 ”
weka . classifiers . functions . SMO −C 1 . 0 −L 0 . 0 0 1 −P 1 . 0 E−12 −N 0 −V −10 −W 1 −K ” weka .classifiers . functions . supportVector . PolyKernel −C 250007 −E 1 . 0 ”
● weka . classifiers . functions . SMO : We are using the SMO class of the Weka jar library as our classifier ● -C 1.0 : Cost parameter for the soft margin SVM classifier. Balances between misclassifications & rigid boundaries. ● -L 0.001 : Tolerance parameter set for SMO convergence. Default ● -P 1.0e-12: Epsilon that determines number of SVs to choose. Larger P => smaller % of SVs
weka . classifiers . functions .