radbastf_snn.m

来自「神经网络的工具箱, 神经网络的工具箱,」· M 代码 · 共 50 行

function f = radbastf_snn(x)
%RADBASTF_SNN Radial basis transfer function.
%
% Syntax
%
%   f = radbastf_snn(x)
%
%    x - input. 
%    f - output for transfer function.
%
% Description
%
%   RADBASTF_SNN is a transfer function. Transfer functions
%   calculate a layer's output from its net input.
%
%   RADBASTF_SNN(x) takes one input,
%     x - net input.
%   and returns 
%     f - exp(-x^2) 
%
%   RADBASTF_SNN(CODE) returns useful information for each CODE string:
%     'deriv'  - Returns name of derivative function.
%
% See also
%
%   DRADBASTF_SNN
%

% This function is a substitute for RADBAS from Matlab's NN Toolbox. 

if nargin < 1, error('Not enough arguments.'); end

if isstr(x)
  switch (x)
    case 'version'
	  f = 3.0;       % <-- Must be 3.0. (can be used in matlabs 
	                 %     Neural network toolbox)
    case 'deriv'
	  f = 'dradbastf_snn'; 
    case 'output'
	  f = [0 1]; 
    case 'active'
	  f = [-2 +2]; 
    otherwise, error('Unrecognized code.')
  end

else
  f = exp(-(x.*x));  
end