深度學習中dropout策略的理解

如今有空整理一下關於深度學習中怎麼加入dropout方法來防止測試過程的過擬合現象。

首先了解一下dropout的實現原理：

這些理論的解釋在百度上有不少。。。。

這裏重點記錄一下怎麼實現這一技術

參考別人的博客，主要http://www.cnblogs.com/dupuleng/articles/4340293.html

講解一下用Matlab中的深度學習工具箱怎麼實現dropout

首先要載入工具包。DeepLearn Toolbox是一個很是有用的matlab deep learning工具包，下載地址：https://github.com/rasmusbergpalm/DeepLearnToolbox

要使用它首先要將該工具包添加到matlab的搜索路徑中，

一、將包複製到matlab 的toolbox中，做者的路徑是D:\program Files\matlab\toolbox\

二、在matlab的命令行中輸入：　　

cd D:\program Files\matlab\toolbox\deepLearnToolbox\
addpath(gepath('D:\program Files\matlab\toolbox\deepLearnToolbox-master\')
savepath   %保存，這樣就不須要每次都添加一次

三、驗證添加是否成功，在命令行中輸入  

which saesetup

果成功就會出現，saesetup.m的路徑D:\program Files\matlab\toolbox\deepLearnToolbox-master\SAE\saesetup.m 

四、使用deepLearnToolbox 工具包，作一個簡單的demo，將autoencoder模型使用dropout先後的結果進行比較。

load mnist_uint8;
train_x = double(train_x(1:2000,:)) / 255;
test_x  = double(test_x(1:1000,:))  / 255;
train_y = double(train_y(1:2000,:));
test_y  = double(test_y(1:1000,:));

%% //實驗一without dropout
rand('state',0)
sae = saesetup([784 100]);
sae.ae{1}.activation_function  = 'sigm';
sae.ae{1}.learningRate         =  1;
opts.numepochs = 10;
opts.batchsize = 100;
sae = saetrain(sae , train_x , opts );
visualize(sae.ae{1}.W{1}(:,2:end)');

nn = nnsetup([784 100 10]);% //初步構造了一個輸入-隱含-輸出層網絡，其中包括了
                           % //權值的初始化，學習率，momentum，激發函數類型，
                           % //懲罰係數，dropout等

nn.W{1} = sae.ae{1}.W{1};
opts.numepochs =  10;   %  //Number of full sweeps through data
opts.batchsize = 100;  %  //Take a mean gradient step over this many samples
[nn, ~] = nntrain(nn, train_x, train_y, opts);
[er, ~] = nntest(nn, test_x, test_y);
str = sprintf('testing error rate is: %f',er);
fprintf(str);

%% //實驗二：with dropout
rand('state',0)
sae = saesetup([784 100]);
sae.ae{1}.activation_function  = 'sigm';
sae.ae{1}.learningRate         =  1;

opts.numepochs = 10;
opts.bachsize = 100;
sae = saetrain(sae , train_x , opts );
figure;
visualize(sae.ae{1}.W{1}(:,2:end)');

nn = nnsetup([784 100 10]);% //初步構造了一個輸入-隱含-輸出層網絡，其中包括了
                           % //權值的初始化，學習率，momentum，激發函數類型，
                           % //懲罰係數，dropout等
nn.dropoutFraction = 0.5;  
nn.W{1} = sae.ae{1}.W{1};
opts.numepochs =  10;   %  //Number of full sweeps through data
opts.batchsize = 100;  %  //Take a mean gradient step over this many samples
[nn, L] = nntrain(nn, train_x, train_y, opts);
[er, bad] = nntest(nn, test_x, test_y);
str = sprintf('testing error rate is: %f',er);
fprintf(str);