在 Keras 中使用深度学习的不同结果

Question

我在tutorial 之后使用 keras 中的深度神经网络进行文本分类，但是当我多次运行以下代码时，我得到了不同的结果。

例如，第一次运行的测试损失为0.88815，第二次运行为0.89030，略高。我想知道随机性从何而来？

import keras
from keras.datasets import reuters


(x_train, y_train), (x_test, y_test) = reuters.load_data(num_words=None, test_split=0.2)
word_index = reuters.get_word_index(path="reuters_word_index.json")



print('# of Training Samples: {}'.format(len(x_train)))
print('# of Test Samples: {}'.format(len(x_test)))

num_classes = max(y_train) + 1
print('# of Classes: {}'.format(num_classes))

index_to_word = {}
for key, value in word_index.items():
    index_to_word[value] = key

print(' '.join([index_to_word[x] for x in x_train[0]]))
print(y_train[0])


from keras.preprocessing.text import Tokenizer

max_words = 10000

tokenizer = Tokenizer(num_words=max_words)
x_train = tokenizer.sequences_to_matrix(x_train, mode='binary')
x_test = tokenizer.sequences_to_matrix(x_test, mode='binary')

y_train = keras.utils.to_categorical(y_train, num_classes)
y_test = keras.utils.to_categorical(y_test, num_classes)


print(x_train[0])
print(len(x_train[0]))

print(y_train[0])
print(len(y_train[0]))


from keras.models import Sequential
from keras.layers import Dense, Dropout, Activation

model = Sequential()
model.add(Dense(512, input_shape=(max_words,)))
model.add(Activation('relu'))
model.add(Dropout(0.5))
model.add(Dense(num_classes))
model.add(Activation('softmax'))



model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
print(model.metrics_names)

batch_size = 32
epochs = 3

history = model.fit(x_train, y_train, batch_size=batch_size, epochs=epochs, verbose=1, validation_split=0.1)
score = model.evaluate(x_test, y_test, batch_size=batch_size, verbose=1)
print('Test loss:', score[0])
print('Test accuracy:', score[1])

Answer 1

我不检查 GPU，但检查 CPU 似乎无法像上面那样使用 Tensorflow 1 作为 Keras 后端来修复种子。因此，我们需要将 Tensorflow 1 更改为 Tensorflow 2。然后，固定种子将起作用。例如，这对我有用。

import os
import numpy as np
import random as rn
import tensorflow as tf

os.environ['PYTHONHASHSEED']= '0'
np.random.seed(1)
rn.seed(1)
tf.set_random_seed(1)

Answer 2

如Keras FAQ中提到的，添加如下代码：

import numpy as np
import tensorflow as tf
import random as rn

# The below is necessary in Python 3.2.3 onwards to
# have reproducible behavior for certain hash-based operations.
# See these references for further details:
# https://docs.python.org/3.4/using/cmdline.html#envvar-PYTHONHASHSEED
# https://github.com/keras-team/keras/issues/2280#issuecomment-306959926

import os
os.environ['PYTHONHASHSEED'] = '0'

# The below is necessary for starting Numpy generated random numbers
# in a well-defined initial state.

np.random.seed(42)

# The below is necessary for starting core Python generated random numbers
# in a well-defined state.

rn.seed(12345)

# Force TensorFlow to use single thread.
# Multiple threads are a potential source of
# non-reproducible results.
# For further details, see: https://stackoverflow.com/questions/42022950/which-seeds have-to-be-set-where-to-realize-100-reproducibility-of-training-res

session_conf = tf.ConfigProto(intra_op_parallelism_threads=1, 
inter_op_parallelism_threads=1)

from keras import backend as K

# The below tf.set_random_seed() will make random number generation
# in the TensorFlow backend have a well-defined initial state.
# For further details, see: https://www.tensorflow.org/api_docs/python/tf/set_random_seed

tf.set_random_seed(1234)

sess = tf.Session(graph=tf.get_default_graph(), config=session_conf)
K.set_session(sess)

# Rest of code follows ...

Answer 3

这是 keras 的常见行为。请参阅 github 的 keras 存储库问题列表中的 this discussion。

例如，在fit function 中，第 9 个参数是关于洗牌的。默认情况下设置为 true。因此，在每个 epoch 中，数据在运行前都会被打乱。这会导致该值每次都发生变化。

设置随机种子会有所帮助。但是，仍然不完全正确。

Answer 4

如果您想每次都获得相同的结果，您需要添加一个随机种子。另见https://machinelearningmastery.com/reproducible-results-neural-networks-keras/。

这可以通过添加：

from numpy.random import seed
seed(42)

如果您使用的是 Tensorflow 后端，您还需要添加：

from tensorflow import set_random_seed
set_random_seed(42)

42 只是一个您可以随意选择的任意数字。这只是随机种子的一个常数，因此您将始终为您的权重获得相同的随机初始化。这将导致给你相同的结果。