# Data scaling
X = (X / 16).astype(np.float32)
# Splitting data
X_train, X_test, Y_train, Y_test = train_test_split(X,
Y,
test_size=0.2,
random_state=0)
# Training
classifier = SupervisedDBNClassification(
hidden_layers_structure=[1000, 1000, 1000],
learning_rate_rbm=0.05,
learning_rate=0.1,
n_epochs_rbm=15,
n_iter_backprop=50,
batch_size=32,
activation_function='relu',
dropout_p=0.2)
classifier.fit(X_train, Y_train)
# Save the model
classifier.save('model.pkl')
# Restore it
classifier = SupervisedDBNClassification.load('model.pkl')
# Test
Y_pred = classifier.predict(X_test)
print 'Done.\nAccuracy: %f' % accuracy_score(Y_test, Y_pred)
#X_scaled_train = preprocessing.scale(X_train)
min_max_scaler = preprocessing.MinMaxScaler()
X_scaled_train = min_max_scaler.fit_transform(X_train)
y_train = train_set[1:, -1]
X_test = test_set[:5000, 1:-1]
#X_scaled_test = preprocessing.scale(X_test)
min_max_scaler = preprocessing.MinMaxScaler()
X_scaled_test = min_max_scaler.fit_transform(X_test)
y_test = test_set[:5000, -1]
# Training
clf = SupervisedDBNClassification(
hidden_layers_structure=[1024, 512],
learning_rate_rbm=0.05,
learning_rate=0.1,
n_epochs_rbm=3,
n_iter_backprop=10,
batch_size=128,
activation_function='sigmoid', # relu->error
dropout_p=0.2)
clf.fit(X_train, y_train)
# Save the model
clf.save('model.pkl')
# Restore it
classifier = SupervisedDBNClassification.load('model.pkl')
# Test
y_pred = classifier.predict(X_test)
print('Done.\nAccuracy: %f' % accuracy_score(y_test, y_pred))