Question

I am going to use scikit SVC with polynomial kernel in the following format: (1 + xTn.xm)^4.
Meaning (one plus the product of xTn.xm) with degree 4. xTn is the xn value that is transposed.
When I look into the scikit documentation they specify the parameters for SVC:

    degree : int, optional (default=3)
 |      Degree of kernel function.
 |      It is significant only in 'poly' and 'sigmoid'.
 |  
 |  gamma : float, optional (default=0.0)
 |      Kernel coefficient for 'rbf' and 'poly'.
 |      If gamma is 0.0 then 1/n_features will be used instead.
 |  
 |  coef0 : float, optional (default=0.0)
 |      Independent term in kernel function.
 |      It is only significant in 'poly' and 'sigmoid'.

The parameter “degree” is relevant and I should use 4.
The parameter coef0 is relevant and I should use the value 1. Is it correct?
The gamma is very unclear in the “Poly” case. Which value should I use for gamma??

Answer 1

From the libsvm documentation: (gamma * u'* v + coef0)^degree (scikit-learn is based on it for the SVC model). A pull-request to improve the documentation is always welcomed 🙂