clear
close all
load('Zemris.mat')
testimages=trainimages;
testlabels=trainlabels;
load('LPALIC_Without_Zemris.mat')


%% Define the Network Layers

layers = [
    imageInputLayer([28 28 1],'Normalization','none','Name','input')
    convolution2dLayer(5,64,'Name','conv1')
    batchNormalizationLayer('Name','BN1')
    reluLayer('Name','relu1')
% 24x24
    convolution2dLayer(5,128,'Name','conv2')
    batchNormalizationLayer('Name','BN2')
    reluLayer('Name','relu2')
% 20x20
    maxPooling2dLayer(2,'Stride',2,'Name','max1')
% 10x10 
    convolution2dLayer(5,176,'Name','conv3')
    batchNormalizationLayer('Name','BN3')
    reluLayer('Name','relu3')
% 6x6 
    convolution2dLayer(5,208,'Name','conv4')
    batchNormalizationLayer('Name','BN4')
    reluLayer('Name','relu4')
% 2x2   
    maxPooling2dLayer(2,'Stride',2,'Name','max2')
  
    fullyConnectedLayer(35,'Name','fc')
    softmaxLayer('Name','softmax')
    classificationLayer('Name','classOutput')];

%% Specify the Training Options

options = trainingOptions('sgdm','LearnRateSchedule','piecewise',...
      'MiniBatchSize',120,'LearnRateDropFactor',0.7,... 
      'LearnRateDropPeriod',2,... 
      'MaxEpochs',10, ...
	'InitialLearnRate',1/40,...
    'Shuffle','every-epoch');  

   convnet = trainNetwork(trainimages,trainlabels,layers,options);

% load('Zemris_result_9790.mat')
   
YTest = classify(convnet,testimages);
accuracy = sum(YTest == testlabels)/numel(testlabels);
Error=(1-accuracy)*100

%% Fine tune with ADAM
lgraph = convnet;
BZ=240;
LR=0.006/BZ;
options = trainingOptions('adam','LearnRateSchedule','piecewise',...
      'MiniBatchSize',BZ,'LearnRateDropFactor',0.95,...
     'verbose',true,...
     'LearnRateDropPeriod',1,... 
      'MaxEpochs',1, ...
	'InitialLearnRate',LR,...
    'Shuffle','every-epoch');  

convnet = trainNetwork(trainimages,lgraph,options);

YTest = classify(convnet,testimages);
accuracy = sum(YTest == testlabels)/numel(testlabels);
Error=(1-accuracy)*100