Contents

Description of demo_binaryclass_mlp.m

Uses perceptrons and multi-layer perceptrons for binary classification with various architectures

clear all
close all
generateData_vert

usage of perceptron binary classification (vert dataset)

options_st = [];
model_st = ml_binaryclass_perceptron(Xtrain, ytrain, options_st);
yhat_st = model_st.predict(model_st, Xtest);
testError_st = mean(yhat_st ~= ytest);
fprintf('Averaged absolute test error with %s is: %.3f\n', ...
        model_st.name, testError_st);

Averaged absolute test error with Perceptron Binary Classification is: 0.418

usage of MLP binary classification (vert dataset)

options_tr = [];
options_tr.nHidden = [3 3 3];
model_tr = ml_binaryclass_MLP(Xtrain, ytrain, options_tr);
yhat_tr = model_tr.predict(model_tr, Xtest);
testError_tr = mean(yhat_tr ~= ytest);
fprintf('Averaged absolute test error with %s is: %.3f\n', ...
        model_tr.name, testError_tr);

Averaged absolute test error with Multi-layer Perceptron Binary Classification is: 0.062

figure;
plot2DClassifier(Xtrain, ytrain, model_st);
figure;
plot2DClassifier(Xtrain, ytrain, model_tr);

generateData_slanted

usage of perceptron binary classification (slanted dataset)

options_st = [];
model_st = ml_binaryclass_perceptron(Xtrain, ytrain, options_st);
yhat_st = model_st.predict(model_st, Xtest);
testError_st = mean(yhat_st ~= ytest);
fprintf('Averaged absolute test error with %s is: %.3f\n', ...
        model_st.name, testError_st);

Averaged absolute test error with Perceptron Binary Classification is: 0.098

usage of MLP binary classification (slanted dataset)

options_tr = [];
options_tr.nHidden = [3 3 3];
model_tr = ml_binaryclass_MLP(Xtrain, ytrain, options_tr);
yhat_tr = model_tr.predict(model_tr, Xtest);
testError_tr = mean(yhat_tr ~= ytest);
fprintf('Averaged absolute test error with %s is: %.3f\n', ...
        model_tr.name, testError_tr);

Averaged absolute test error with Multi-layer Perceptron Binary Classification is: 0.156

figure;
plot2DClassifier(Xtrain, ytrain, model_st);
figure;
plot2DClassifier(Xtrain, ytrain, model_tr);

generateData_groups

usage of perceptron binary classification (groups dataset)

options_st = [];
model_st = ml_binaryclass_perceptron(Xtrain, ytrain, options_st);
yhat_st = model_st.predict(model_st, Xtest);
testError_st = mean(yhat_st ~= ytest);
fprintf('Averaged absolute test error with %s is: %.3f\n', ...
        model_st.name, testError_st);

Averaged absolute test error with Perceptron Binary Classification is: 0.591

usage of MLP binary classification (groups dataset)

options_tr = [];
options_tr.nHidden = [5 5 5];
model_tr = ml_binaryclass_MLP(Xtrain, ytrain, options_tr);
yhat_tr = model_tr.predict(model_tr, Xtest);
testError_tr = mean(yhat_tr ~= ytest);
fprintf('Averaged absolute test error with %s is: %.3f\n', ...
        model_tr.name, testError_tr);

Averaged absolute test error with Multi-layer Perceptron Binary Classification is: 0.120

figure;
plot2DClassifier(Xtrain, ytrain, model_st);
figure;
plot2DClassifier(Xtrain, ytrain, model_tr);

Published with MATLAB® R2015a