project1.m~

% cd('/home/binroot/tutorial/code/examples');
pathTemp = genpath('./');
addpath(pathTemp);
% setup;

% task 1
% cnn_cifar;
% cd('data/cifar-lenet/');
% load('net-epoch-30.mat');
% disp(info.train.error);
% disp(info.val.error);
% t = 1:30;
% plot(t,info.train.error);
% hold on
% plot(t,info.val.error);
% title('training & validation error');
% legend('train','val');

% task 2
% net = cnn_cifar_init();
% cnn_cifar;
load('net-epoch-30-blk1.mat');
disp(info.train.error);
disp(info.val.error);
t = 1:30;
plot(t,info.train.error);
hold on
plot(t,info.val.error);
title('training & validation error');
legend('train','val');
xlabel('Epoch');
ylabel('Error');
xlim([0,1]);
ylim()
% copyfile model.mat model-blk3.mat;
% copyfile data/cifar-lenet/net-epoch-30.mat data/cifar-lenet/net-epoch-30-blk3.mat;

% task 3
% close all
% model = load('model-full.mat');
% model.net.layers = model.net.layers(1:end-1); % remove softmax
% for i = 1:10
%     img = imread(['images/',num2str(i),'.png']);
%     img = single(img) - model.net.averageImage;
%     res = vl_simplenn(model.net,img);
%     response = res(2).x;
%     count = 1;
%     figure;
%     for j = 1:32
%         subplot(4,8,count);
%         image(response(:,:,j));
%         axis tight;
%         axis off;
%         daspect([1 1 1]);
%         count = count + 1;
%     end
%     print(['filter_response/',num2str(i)],'-djpeg');
% zend