text classification with RNN

The purpose of this tutorial is to lead you to learn to use RNN Classify text

The data set used this time is IMDB, Altogether 50000 Movie reviews , among 25000 It's a training set , in addition 25000 This is the test set

First we need to load the dataset , Can pass TFDS It's easy to download the data set , As shown in the following code

dataset, info = tfds.load('imdb_reviews', with_info=True, as_supervised=True)
​
train_dataset, test_dataset = dataset['train'], dataset['test']
​
train_dataset.element_spec

Next we need to create text encoder, Can pass tf.keras.layers.experimental.preprocessing.TextVectorization Realization , As shown in the following code

VOCAB_SIZE = 1000
encoder = tf.keras.layers.experimental.preprocessing.TextVectorization(
    max_tokens=VOCAB_SIZE
)
encoder.adapt(train_dataset.map(lambda text, label: text))

Next we need to build a model , The following figure is the model structure diagram

The corresponding code is as follows

model = tf.keras.Sequential([
    encoder,
    tf.keras.layers.Embedding(
        input_dim=len(encoder.get_vocabulary()),
        output_dim=64,
        # Use masking to handle the variable sequence lengths
        mask_zero=True),
    tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(64)),
    tf.keras.layers.Dense(64, activation='relu'),
    tf.keras.layers.Dense(1)
])
​
model.compile(loss=tf.keras.losses.BinaryCrossentropy(from_logits=True),
              optimizer=tf.keras.optimizers.Adam(1e-4),
              metrics=['accuracy'])

To this step , We can start training , And model evaluation after training

history = model.fit(train_dataset, epochs=10,
                    validation_data=test_dataset,
                    validation_steps=30)
​
test_loss, test_acc = model.evaluate(test_dataset)
​
print('Test Loss:', test_loss)
print('Test Accuracy:', test_acc)

The above is the record of training results

Code address : https://codechina.csdn.net/csdn_codechina/enterprise_technology/-/blob/master/text_classification_rnn.ipynb