SVM parameter tunning

Hi,

I am trying to optimize my parameters for SVM algorithm. As I am new to Knime and ML

, I am struggling to find the best approach. In rbf kernel, there is sigma parameter instead of gamma and C. what does it correspond and in which range should I analyse it (grid search)? will also be nice to know parameter ranges for other kernels.

finally which search strategy will be the best in the parameter optimization loop.

Welcome to the KNIME forum! Thank you for your question. Here is a more detailed documentation about SVMs: https://www.csie.ntu.edu.tw/~cjlin/papers/guide/guide.pdf
In addition, you can find an example workflow with SVM and Parameter Optimization here: https://hub.knime.com/knime/spaces/Examples/latest/04_Analytics/11_Optimization/07_Cross_Validation_with_SVM_and_Parameter_Optimization. In this workflow, the sigma parameter for the SVM is varied and controlled by a flow variable.

Hope that helps,
Jeany

3 Likes

This topic was automatically closed 182 days after the last reply. New replies are no longer allowed.