ModErn Text Analysis
META Enumerates Textual Applications
Public Member Functions | Private Attributes | List of all members
meta::classify::kernel::sigmoid Class Reference

A sigmoid kernel function for a linear classifier to adapt it to data that is not linearly separable. More...

#include <sigmoid.h>

Public Member Functions

 sigmoid (double alpha, double c)
 Constructs a new sigmoid kernel with the given parameters. More...
 
template<class PostingsData >
double operator() (const PostingsData &first, const PostingsData &second) const
 Computes the value of \(K(first, second)\).
 

Private Attributes

double alpha_
 \(\alpha\), the coefficient for the dot product.
 
double c_
 \(c\), the additional scalar term
 
polynomial dot_ {1, 0.0}
 Internal "helper" kernel.
 

Detailed Description

A sigmoid kernel function for a linear classifier to adapt it to data that is not linearly separable.

Uses the general form of: \(K(x,y) = \tanh(\alpha x^T y + c)\)

Constructor & Destructor Documentation

meta::classify::kernel::sigmoid::sigmoid ( double  alpha,
double  c 
)
inline

Constructs a new sigmoid kernel with the given parameters.

Parameters
alpha\(\alpha\), the coefficient for the dot product
c\(c\), the additional scalar term

The documentation for this class was generated from the following files: