tensorflow2.x RNN进行情感分析

使用RNN(循环神经网络)对电影评论进行情感分析，结果为positive或negative，分别代表积极和消极的评论。至于为什么使用RNN而不是普通的前馈神经网络，是因为RNN能够存储序列单词信息，得到的结果更为准确。
使用的RNN模型架构如下

上代码：

import  os
import  tensorflow as tf
import  numpy as np
from tensorflow import keras


"""在这里我们将使用RNN(循环神经网络)对电影评论进行情感分析，结果为positive或negative，分别代表积极和消极的评论。
至于为什么使用RNN而不是普通的前馈神经网络，是因为RNN能够存储序列单词信息，得到的结果更为准确。这里我们将使用一个带有标签的影评数据集进行训练模型。"""


tf.random.set_seed(22)
np.random.seed(22)
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
assert tf.__version__.startswith('2.')



# fix random seed for reproducibility
np.random.seed(7)
#加载数据集，但仅保留前10000个字，其余为零
top_words = 10000
# 最长填充序列
max_review_length = 80
(X_train, y_train), (X_test, y_test) = keras.datasets.imdb.load_data('F:/PycharmProjects/TensorFlow-2.0-Tutorials/09-RNN-Sentiment-Analysis/data/imdb.npz', num_words=top_words)
# X_train = tf.convert_to_tensor(X_train)
# y_train = tf.one_hot(y_train, depth=2)

#将序列填充到相同的长度:80。
x_train = keras.preprocessing.sequence.pad_sequences(X_train, maxlen=max_review_length)
x_test = keras.preprocessing.sequence.pad_sequences(X_test, maxlen=max_review_length)
print('x_train shape:', x_train.shape)
print('x_test shape:', x_test.shape)


class RNN(keras.Model):

    def __init__(self, units, num_classes, num_layers):
        super(RNN, self).__init__()


        # self.cells = [keras.layers.LSTMCell(units) for _ in range(num_layers)]
        #
        # self.rnn = keras.layers.RNN(self.cells, unroll=True)
        self.rnn = keras.layers.LSTM(units, return_sequences=True)
        self.rnn2 = keras.layers.LSTM(units)

        # self.cells = (keras.layers.LSTMCell(units) for _ in range(num_layers))
        # #
        # self.rnn = keras.layers.RNN(self.cells, return_sequences=True, return_state=True)
        # self.rnn = keras.layers.LSTM(units, unroll=True)
        # self.rnn = keras.layers.StackedRNNCells(self.cells)


        # have 1000 words totally, every word will be embedding into 100 length vector
        # the max sentence lenght is 80 words
        self.embedding = keras.layers.Embedding(top_words, 100, input_length=max_review_length)
        self.fc = keras.layers.Dense(1)

    def call(self, inputs, training=None, mask=None):

        print('x', inputs.shape)
        """首先，将单词传入embedding层，之所以使用嵌入层，是因为单词数量太多，使用嵌入式方式词向量来表示单词更有效率"""
        # [b, sentence len] => [b, sentence len, word embedding]
        x = self.embedding(inputs)
        """其次，通过embedding层，新的单词表示传入LSTM cells。这将是一个递归链接网络，所以单词的序列信息会在网络之间传递"""
        x = self.rnn(x) 
        x = self.rnn2(x) 
        # print('rnn', x.shape)
        """最后，LSTM cells连接一个sigmoid output layer。使用sigmoid可以预测该文本是积极的还是消极的情感"""
        x = self.fc(x)
        print(x.shape)

        return x


def main():

    units = 64
    num_classes = 2
    batch_size = 32
    epochs = 5

    model = RNN(units, num_classes, num_layers=2)


    model.compile(optimizer=keras.optimizers.Adam(0.001),
                  loss=keras.losses.BinaryCrossentropy(from_logits=True),
                  metrics=['accuracy'])
    """fit()用于使用给定输入训练模型."""
    model.fit(x_train, y_train, batch_size=batch_size, epochs=epochs,
              validation_data=(x_test, y_test), verbose=1)

    """model.predict只返回y_pred"""
    out = model.predict(x_train)
    print("out:", out)
    """evaluate用于评估您训练的模型。它的输出是准确度或损失，而不是对输入数据的预测。"""
    scores = model.evaluate(x_test, y_test, batch_size, verbose=1)
    print("Final test loss and accuracy :", scores)




if __name__ == '__main__':
    main()