def train_student(
student,
teacher,
):
student.evaluate(test_dataset_x[0], feature_test_data, verbose=2)
student_history = student.fit(train_dataset_x[0],
feature_train_data,
batch_size=32,
epochs=epochs,
validation_data=(test_dataset_x[0],
feature_test_data),
verbose=verbose,
callbacks=[model_checkpoint_callback])
print('{} train:'.format(student.name),
student_history.history['mean_squared_error'])
print('{} test:'.format(student.name),
student_history.history['val_mean_squared_error'])
dataset_update(student_history, student, parameters, name='sample_5')
print(student.get_weights()[3][:4, :4])
save_model(student, save_model_path, parameters, checkpoint=False)
student.load_weights(checkpoint_filepath)
save_model(student, save_model_path, parameters, checkpoint=True)
student.evaluate(test_dataset_x[0], feature_test_data, verbose=2)
student_predictions = student(test_dataset_x[0])
train_predictions = student(train_dataset_x[0])
return student_predictions, train_predictions
def train_student(
student,
teacher,
):
student.evaluate(test_dataset_x[0], feature_test_data, verbose=2)
student_history = student.fit(train_dataset_x[0],
feature_train_data,
batch_size=32,
epochs=epochs,
validation_data=(test_dataset_x[0],
feature_test_data),
verbose=verbose,
callbacks=[model_checkpoint_callback])
print('{} train:'.format(student.name),
student_history.history['mean_squared_error'])
print('{} test:'.format(student.name),
student_history.history['val_mean_squared_error'])
print(student.get_weights()[3][:4, :4])
save_model(student, save_model_path, parameters, checkpoint=False)
student.load_weights(checkpoint_filepath)
save_model(student, save_model_path, parameters, checkpoint=True)
student.evaluate(test_dataset_x[0], feature_test_data, verbose=2)
student_test_data = np.zeros([5000, 8, 8, num_feature])
student_train_data = np.zeros([45000, 8, 8, num_feature])
start = 0
end = batch_size
count = 0
for batch in range(len(train_dataset_x[0]) // batch_size + 1):
count += 1
train_temp = student(train_dataset_x[0][start:end]).numpy()
student_train_data[start:end, :, :, :] = train_temp[:, :, :, :]
start += batch_size
end += batch_size
start = 0
end = batch_size
count = 0
for batch in range(len(test_dataset_x[0]) // batch_size + 1):
count += 1
test_temp = student(test_dataset_x[0][start:end]).numpy()
student_test_data[start:end, :, :, :] = test_temp[:, :, :, :]
start += batch_size
end += batch_size
#Evaluate per feature
var_list = []
for feature_indx in range(num_feature):
var = np.var(student_test_data[:, :, :, feature_indx] -
feature_test_data[:, :, :, feature_indx])
var_list.append(var)
parameters['student_var'] = var_list
dataset_update(student_history,
student,
parameters,
name='multi_features_{}'.format(num_feature))
return student_test_data, student_train_data
train_teacher_data,
batch_size=32,
epochs=epochs,
validation_data=(test_dataset_x[0],
test_teacher_data),
verbose=verbose)
print('student3 net train:', student_history3.history['mean_squared_error'])
print('student3 net test:', student_history3.history['val_mean_squared_error'])
now = datetime.now()
print(now.strftime("%d/%m/%Y %H:%M:%S"))
student3_prediction, teacher_target = feature_learning3.get_predictions(
[test_dataset_x[0][50:150], test_teacher_data[50:150]]) * 1
# write_to_file(student_history3, student_net3,parameters)
now = datetime.now()
print(now.strftime("%d/%m/%Y %H:%M:%S"))
dataset_update(student_history3, student_net3, parameters)
now = datetime.now()
print(now.strftime("%d/%m/%Y %H:%M:%S"))
teacher_mean, teacher_std = feature_learning3.get_teacher_std(
test_teacher_data)
parameters['teacher_mean'] = teacher_mean
parameters['teacher_var'] = teacher_std**2
print('teacher feature mean =', teacher_mean, 'teacher feature var =',
teacher_std**2)
print(np.std(target))
print(np.std(student3_prediction))
with open(
prediction_data_path +