def accuracy(y_true, y_pred):
'''
Temporary metrics to overcome "from_categorical" missing in standard metrics
'''
truth = from_categorical(y_true)
predicted = from_categorical(y_pred)
return mean_accuracy_score(truth, predicted)
def test_mean_accuracy_score(self, size):
y_true = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
y_pred = np.random.choice([0., 1.], p=[.5, .5], size=(size, ))
metric = tf.keras.metrics.Accuracy()
res_py = mean_accuracy_score(y_true, y_pred)
metric.update_state(y_true, y_pred)
res_tf = metric.result().numpy()
np.testing.assert_allclose(res_tf, res_py, atol=1e-8, rtol=1e-5)
# model.add(Cost_layer(cost_type=cost_type.mse))
# model.compile(optimizer=SGD(lr=0.01, decay=0., lr_min=0., lr_max=np.inf))
model.compile(optimizer=Adam(), metrics=[accuracy])
print('*************************************')
print('\n Total input dimension: {}'.format(X_train.shape), '\n')
print('**************MODEL SUMMARY***********')
model.summary()
print('\n***********START TRAINING***********\n')
# Fit the model on the training set
model.fit(X=X_train, y=y_train, max_iter=10, verbose=True)
print('\n***********START TESTING**************\n')
# Test the prediction with timing
loss, out = model.evaluate(X=X_test, truth=y_test, verbose=True)
truth = from_categorical(y_test)
predicted = from_categorical(out)
accuracy = mean_accuracy_score(truth, predicted)
print('\nLoss Score: {:.3f}'.format(loss))
print('Accuracy Score: {:.3f}'.format(accuracy))
# SGD : best score I could obtain was 94% with 10 epochs, lr = 0.01 %
# Momentum : best score I could obtain was 93% with 10 epochs
# Adam : best score I could obtain was 95% with 10 epochs