52 lines
1.3 KiB
Mathematica
52 lines
1.3 KiB
Mathematica
|
%--------------------------------------------------------------------------
|
||
|
|
% ENSEEIHT - 1SN - Analyse de donnees
|
||
|
|
% TP4 - Reconnaissance de chiffres manuscrits par k plus proches voisins
|
||
|
|
% TP4AD.m
|
||
|
|
%--------------------------------------------------------------------------
|
||
|
|
|
||
|
|
clear
|
||
|
|
close all
|
||
|
|
clc
|
||
|
|
|
||
|
|
% Chargement des images d'apprentissage et de test
|
||
|
|
load MNIST
|
||
|
|
|
||
|
|
% database_test_images 10000x784
|
||
|
|
% database_test_labels 10000x1
|
||
|
|
% database_train_images 60000x784
|
||
|
|
% database_train_labels 60000x1
|
||
|
|
|
||
|
|
DataA = database_train_images;
|
||
|
|
clear database_train_images
|
||
|
|
DataT = database_test_images;
|
||
|
|
clear database_test_images
|
||
|
|
labelA = database_train_labels;
|
||
|
|
clear database_train_labels
|
||
|
|
labelT = database_test_labels;
|
||
|
|
clear database_test_labels
|
||
|
|
|
||
|
|
% Choix du nombre de voisins
|
||
|
|
K = 2;
|
||
|
|
|
||
|
|
% Initialisation du vecteur des classes
|
||
|
|
ListeClass = 0:9;
|
||
|
|
|
||
|
|
% Nombre d'images test
|
||
|
|
[Nt,~] = size(DataT);
|
||
|
|
Nt_test = Nt/1000; % A changer, pouvant aller de 1 jusqu'à Nt
|
||
|
|
|
||
|
|
% Classement par l'algorithme des k-ppv
|
||
|
|
[Partition] = kppv(DataA, labelA, DataT, Nt_test, K, ListeClass);
|
||
|
|
|
||
|
|
% affichage des images avec leur label calculé
|
||
|
|
for i=1:6
|
||
|
|
|
||
|
|
subplot(2, 3, i);
|
||
|
|
img = reshape(DataT(i,:), 28, 28);
|
||
|
|
imagesc(img);
|
||
|
|
colormap gray;
|
||
|
|
axis image off;
|
||
|
|
title(num2str(Partition(i)));
|
||
|
|
|
||
|
|
end
|