前提・実現したいこと

hyperoptでDNNを最適化したいです。でもデータはfunctionの形ではなくarrayで代入したいです。

発生している問題・エラーメッセージ

Unexpected error: <class 'IndentationError'>
Traceback (most recent call last):
  File "/Users/qingfengliu/Dropbox/Feng_and_Liu/DMC/DMC_cluster/output/dnn_working.py", line 111, in <module>
    Loss_test = dnn_regression(data, Random_seed, Sample_size)
  File "/Users/qingfengliu/Dropbox/Feng_and_Liu/DMC/DMC_cluster/output/dnn_working.py", line 97, in dnn_regression
    trials=Trials())
  File "/Users/qingfengliu/.conda/envs/Example/lib/python3.7/site-packages/hyperas/optim.py", line 69, in minimize
    keep_temp=keep_temp)
  File "/Users/qingfengliu/.conda/envs/Example/lib/python3.7/site-packages/hyperas/optim.py", line 106, in base_minimizer
    from temp_model import keras_fmin_fnct, get_space
  File "/Users/qingfengliu/Dropbox/Feng_and_Liu/DMC/DMC_cluster/output/temp_model.py", line 75
    x_train, x_val, x_test, y_train, y_val, y_test = dgp(self.random_seed, self.sample_size)
    ^
IndentationError: unexpected indent

該当のソースコード

import os

import tensorflow as tf
import numpy as np

os.environ["KERAS_BACKEND"] = "plaidml.keras.backend"

from tensorflow import keras

import numpy as np

from keras.utils import to_categorical
from sklearn.model_selection import train_test_split
from keras.layers.advanced_activations import PReLU
from sklearn.preprocessing import MinMaxScaler
from hyperopt import Trials, STATUS_OK, tpe
from keras.layers.core import Dense, Dropout, Activation
from keras.models import Sequential
from hyperas import optim
from hyperas.distributions import choice, uniform
from tensorflow.keras.mixed_precision import experimental as mixed_precision
from simulation import dgp

class ImportData:
def init(self, random_seed, sample_size):
# Hold this implementation specific arguments as the fields of the class.
self.random_seed = random_seed
self.sample_size = sample_size

def data(self):
    # Calculate an objective value by using the extra arguments.
    x_train, x_val, x_test, y_train, y_val, y_test = dgp(self.random_seed, self.sample_size)
    return np.concatenate([x_train, x_val]), np.concatenate([y_train, y_val]), x_test, y_test

def create_regression_model(x_train, y_train):
model = Sequential()
model.add(Dense({{choice([50, 100, 1000])}}, input_shape=(len(x_train[0]),)))
model.add(Activation({{choice(['relu', 'tanh'])}}))
model.add(Dropout({{uniform(0, 1)}}))
model.add(Dense({{choice([50, 100, 1000])}}))
model.add(Activation({{choice(['relu', 'tanh'])}}))
model.add(Dropout({{uniform(0, 1)}}))
model.add(Dense({{choice([50, 100, 1000])}}))
model.add(Activation('relu'))
model.add(Dropout({{uniform(0, 1)}}))
model.add(Dense({{choice([50, 100, 1000])}}))
model.add(Activation('relu'))
model.add(Dropout({{uniform(0, 1)}}))
model.add(Dense({{choice([50, 100, 1000])}}))
model.add(Activation('relu'))
model.add(Dropout({{uniform(0, 1)}}))
model.add(Dense({{choice([50, 100, 1000])}}))
model.add(Dense(1))
# model.add(Activation('softmax', dtype='float32')) # mixed_float16

model.compile(loss='mean_squared_error', metrics=['accuracy'],
              optimizer={{choice(['rmsprop', 'adam', 'sgd'])}})

result = model.fit(x_train, y_train,
                   batch_size={{choice([50, 100, 1000, 5000])}},
                   epochs={{choice([10, 20, 30, 50, 100])}},
                   verbose=2,
                   validation_split=0.1)

validation_loss = np.amin(result.history['val_loss'])
# print('Best validation acc of epoch:', validation_acc)
return {'loss': validation_loss, 'status': STATUS_OK, 'model': model}

def dnn_regression(data_fun, random_seed, sample_size):
# mixed_float16
policy = mixed_precision.Policy('mixed_float16')
mixed_precision.set_policy(policy)
best_run, best_model = optim.minimize(model=create_regression_model,
data=data_fun,
algo=tpe.suggest,
max_evals=1,
trials=Trials())
x_train, y_train, x_test, y_test = ImportData(random_seed, sample_size).data()
# evacuation of best performing model
loss_test, acc_test = best_model.evaluate(x_test, y_test)
best_model.summary()
return loss_test

if name == 'main':
# print(data_set)
Random_seed, Sample_size = 1, 100
data = ImportData(Random_seed, Sample_size)
data = data.data
# print(data())
Loss_test = dnn_regression(data, Random_seed, Sample_size)
print(Loss_test)