reinitweights_snn.m

来自「神经网络的工具箱, 神经网络的工具箱,」· M 代码 · 共 29 行

function net = reinitweights_snn(net, std)
%REINITWEIGHTS_SNN (re)initialize connection weights and biases.
%
% Syntax
%
%  net = reinitweights_snn(net, std)
%
% Description
%
%  REINITWEIGHTS_SNN sets the connection weights and biases at random
%                    normal variables with mean zero.
%
%  REINITWEIGHTS_SNN takes
%     net   - a net_struct.
%     std   - standard deviation for connection weights and biases.
%  and returns 
%     net   - a net_struct with the new connection weights and biases.
%

if (nargin <  2)
   std = 0.01;
end

for i = 1:net.numLayers
    net.biases{i} = std*randn(size(net.biases{i}));
    net.weights{i} = std*randn(size(net.weights{i}));
end