model_checkpoint_callback = ModelCheckpoint(filepath=checkpoint_filepath,
monitor='val_loss',
save_best_only=True,
mode='auto')
model.compile(optimizer=adam,
loss='categorical_crossentropy',
metrics=[
'accuracy',
tf.keras.metrics.CategoricalAccuracy(), Evaluator.precision,
Evaluator.recall, Evaluator.fmeasure
])
history = model.fit(x_train,
y_train,
epochs=epochs,
batch_size=batch_size,
validation_split=0.11,
callbacks=[model_checkpoint_callback])
model.load_weights(checkpoint_filepath) # Error in Windows environment
# Validation curves
Evaluator.plot_validation_curves(model_name, history)
y_pred = model.predict(x_test, batch_size=64)
# Experiment result
Evaluator.calculate_measure(model, x_test, y_test)
Evaluator.plot_confusion_matrix(model_name, y_test, y_pred)
Evaluator.plot_roc_curves(model_name, y_test, y_pred)
loss='categorical_crossentropy',
metrics=[
'accuracy',
tf.keras.metrics.CategoricalAccuracy(),
Evaluator.precision, Evaluator.recall,
Evaluator.fmeasure
])
dt_start_predict = datetime.now()
y_pred = best_model.predict(x_test, batch_size=64)
dt_end_predict = datetime.now()
# Validation curves
Evaluator.plot_validation_curves(model_name, history)
Evaluator.print_validation_report(history)
# Experimental result
Evaluator.calculate_measure(best_model, x_test, y_test)
# Save confusion matrix
Evaluator.plot_confusion_matrix(model_name,
y_test,
y_pred,
title='Confusion matrix',
normalize=True)
# Print Training and predicting time
print('Train time: ' + str((dt_end_train - dt_start_train)))
print('Predict time: ' + str((dt_end_predict - dt_start_predict)))
''' Predict phrase '''
best_model = cnn_bilstm_att()
best_model.load_weights('./trained_models/' + model_name+ '.hdf5')
best_model.compile(optimizer=adam, loss='binary_crossentropy',
metrics=['accuracy', tf.keras.metrics.BinaryAccuracy(),
Evaluator.precision, Evaluator.recall, Evaluator.fmeasure])
dt_start_predict = datetime.now()
y_pred = best_model.predict(x_test, batch_size=64)
dt_end_predict = datetime.now()
# Validation curves
Evaluator.plot_validation_curves(model_name, history, type='binary')
Evaluator.print_validation_report(history)
# Experimental result
result = best_model.evaluate(x_test, y_test, batch_size=64)
result_dic = dict(zip(best_model.metrics_names, result))
print('\nAccuracy: {}\n Binary_Accuracy: {}\n'
'Precision: {}\nRecall: {}\n F-1Score {}\n'
.format(result_dic['accuracy'], result_dic['binary_accuracy'],
result_dic['precision'], result_dic['recall'], result_dic['fmeasure']))
Evaluator.calculate_measure_binary(best_model, x_test, y_test)
# Save confusion matrix
Evaluator.plot_confusion_matrix(model_name, y_test, y_pred, title='Confusion matrix', normalize=False, classes=[0,1])