原文地址：http://blog.csdn.net/superdont/article/details/5497654

注意：rands函数与rand不同

有如下程序，

输出如下：

a =

   -0.5749
    0.6785
    0.2576
   -0.7325
   -0.5857


b =

    0.2144
    0.2598
   -0.2590
    0.1503
   -0.0972


c =

   -0.9122   -0.9456   -0.3746   -0.9743   -0.2321


d =

    0.3662

有关rands函数说明如下：

rands 

Symmetric random weight/bias initialization function 
Syntax
W = rands(S,PR)
M = rands(S,R)
v = rands(S);

Description

rands is a weight/bias initialization function. 
rands(S,PR) takes, 
S -- Number of neurons 
PR -- R x 2 matrix of R input ranges 
and returns an S-by-R weight matrix of random values between -1 and 1. 
rands(S,R) returns an S-by-R matrix of random values. rands(S) returns an S-by-1 vector of random values. 
Examples

Here three sets of random values are generated with rands. rands(4,[0 1; -2 2])
rands(4)
rands(2,3)

Network Use

To prepare the weights and the bias of layer i of a custom network to be initialized with rands Set net.initFcn to 'initlay'. (net.initParam will automatically become initlay's default parameters.) Set net.layers{i}.initFcn to 'initwb'. Set each net.inputWeights{i,j}.initFcn to 'rands'. Set each net.layerWeights{i,j}.initFcn to 'rands'. Set each net.biases{i}.initFcn to 'rands'. 
To initialize the network call init

没有说明，如何出现上述结果，百度之：

function [w,b] = rands(s,pr)



%RANDS Symmetric random weight/bias initialization function.



%



%  Syntax



%



%    W = rands(S,PR)



%    M = rands(S,R)



%    v = rands(S);



%



%  Description



%



%    RANDS is a weight/bias initialization function.



%



%    RANDS(S,PR) takes,



%      S  - number of neurons.



%      PR - Rx2 matrix of R input ranges.



%    and returns an S-by-R weight matrix of random values between -1 and 1.



%



%    RANDS(S,R) returns an S-by-R matrix of random values.



%    RANDS(S) returns an S-by-1 vector of random values.



%



%  Examples



%



%    Here three sets of random values are generated with RANDS.



%



%      rands(4,[0 1; -2 2])



%      rands(4)



%      rands(2,3)



%



%  Network Use



%



%    To prepare the weights and the bias of layer i of a custom network



%    to be initialized with RANDS:



%    1) Set NET.initFcn to 'initlay'.



%       (NET.initParam will automatically become INITLAY's default parameters.)



%    2) Set NET.layers{i}.initFcn to 'initwb'.



%    3) Set each NET.inputWeights{i,j}.initFcn to 'rands'.



%       Set each NET.layerWeights{i,j}.initFcn to 'rands';



%       Set each NET.biases{i}.initFcn to 'rands'.



%



%    To initialize the network call INIT.



%



%  See also RANDNR, RANDNC, INITWB, INITLAY, INIT







% Mark Beale, 1-31-92



% Revised 12-15-93, MB



% Revised 11-31-97, MB



% Copyright 1992-2008 The MathWorks, Inc.



% $Revision: 1.1.6.7 $  $Date: 2008/06/20 08:04:33 $







fn = mfilename;



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



if ischar(s)



  switch(s)



    case 'name'



      w = 'Midpoint';



    otherwise, error('NNET:Arguments',['Unrecognized code: ''' s ''''])



  end



  return



end







if nargin == 1



  r = 1;



elseif size(pr,2) == 1



  r = pr;



else



  r = size(pr,1);



end



w = 2*rand(s,r)-1;//注意










% **[ NNT2 Support ]**



if nargout == 2