Novel Feature Selection Method for Nonlinear Support Vector Regression

Algorithm 1

Feature selection for nonlinear support vector regression.

	Input: Training set . The appropriate kernel parameter , parameter C;
	Output: , and ;
	Begin
	Start with ; set iteration number ;
	whiledo
	Find the solution to problem (15) with the fixed ; Compute each feature score by (15).
	for i = 1 : n
	Ifthen
	else
	end
	end
	Ifthen
	Converged and output and as the final solutions.
	else
	set
	end
	end
	Output and as the final solutions;
	end