Text classification with an RNN

View on TensorFlow.org Run in Google Colab View source on GitHub Download notebook

This text classification tutorial trains a recurrent neural network on the IMDB large movie review dataset for sentiment analysis.

from __future__ import absolute_import, division, print_function, unicode_literals

!pip install -q tensorflow-gpu==2.0.0-beta1
import tensorflow_datasets as tfds
import tensorflow as tf

Import matplotlib and create a helper function to plot graphs:

import matplotlib.pyplot as plt


def plot_graphs(history, string):
  plt.plot(history.history[string])
  plt.plot(history.history['val_'+string])
  plt.xlabel("Epochs")
  plt.ylabel(string)
  plt.legend([string, 'val_'+string])
  plt.show()

Setup input pipeline

The IMDB large movie review dataset is a binary classification dataset—all the reviews have either a positive or negative sentiment.

Download the dataset using TFDS. The dataset comes with an inbuilt subword tokenizer.

dataset, info = tfds.load('imdb_reviews/subwords8k', with_info=True,
                          as_supervised=True)
train_dataset, test_dataset = dataset['train'], dataset['test']
Downloading and preparing dataset imdb_reviews (80.23 MiB) to /home/kbuilder/tensorflow_datasets/imdb_reviews/subwords8k/0.1.0...

HBox(children=(IntProgress(value=1, bar_style='info', description='Dl Completed...', max=1, style=ProgressStyl…
HBox(children=(IntProgress(value=1, bar_style='info', description='Dl Size...', max=1, style=ProgressStyle(des…




HBox(children=(IntProgress(value=1, bar_style='info', max=1), HTML(value='')))


HBox(children=(IntProgress(value=0, description='Shuffling...', max=10, style=ProgressStyle(description_width=…
WARNING: Logging before flag parsing goes to stderr.
W0614 18:49:21.070619 139970744243968 deprecation.py:323] From /home/kbuilder/.local/lib/python3.5/site-packages/tensorflow_datasets/core/file_format_adapter.py:209: tf_record_iterator (from tensorflow.python.lib.io.tf_record) is deprecated and will be removed in a future version.
Instructions for updating:
Use eager execution and: 
`tf.data.TFRecordDataset(path)`

HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…


HBox(children=(IntProgress(value=1, bar_style='info', max=1), HTML(value='')))


HBox(children=(IntProgress(value=0, description='Shuffling...', max=10, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…


HBox(children=(IntProgress(value=1, bar_style='info', max=1), HTML(value='')))


HBox(children=(IntProgress(value=0, description='Shuffling...', max=20, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
HBox(children=(IntProgress(value=1, bar_style='info', description='Reading...', max=1, style=ProgressStyle(des…
HBox(children=(IntProgress(value=0, description='Writing...', max=2500, style=ProgressStyle(description_width=…
Dataset imdb_reviews downloaded and prepared to /home/kbuilder/tensorflow_datasets/imdb_reviews/subwords8k/0.1.0. Subsequent calls will reuse this data.

As this is a subwords tokenizer, it can be passed any string and the tokenizer will tokenize it.

tokenizer = info.features['text'].encoder
print ('Vocabulary size: {}'.format(tokenizer.vocab_size))
Vocabulary size: 8185
sample_string = 'TensorFlow is cool.'

tokenized_string = tokenizer.encode(sample_string)
print ('Tokenized string is {}'.format(tokenized_string))

original_string = tokenizer.decode(tokenized_string)
print ('The original string: {}'.format(original_string))

assert original_string == sample_string
Tokenized string is [6307, 2327, 4043, 4265, 9, 2724, 7975]
The original string: TensorFlow is cool.

The tokenizer encodes the string by breaking it into subwords if the word is not in its dictionary.

for ts in tokenized_string:
  print ('{} ----> {}'.format(ts, tokenizer.decode([ts])))
6307 ----> Ten
2327 ----> sor
4043 ----> Fl
4265 ----> ow 
9 ----> is 
2724 ----> cool
7975 ----> .
BUFFER_SIZE = 10000
BATCH_SIZE = 64
train_dataset = train_dataset.shuffle(BUFFER_SIZE)
train_dataset = train_dataset.padded_batch(BATCH_SIZE, train_dataset.output_shapes)

test_dataset = test_dataset.padded_batch(BATCH_SIZE, test_dataset.output_shapes)

Create the model

Build a tf.keras.Sequential model and start with an embedding layer. An embedding layer stores one vector per word. When called, it converts the sequences of word indices to sequences of vectors. These vectors are trainable. After training (on enough data), words with similar meanings often have similar vectors.

This index-lookup is much more efficient than the equivalent operation of passing a one-hot encoded vector through a tf.keras.layers.Dense layer.

A recurrent neural network (RNN) processes sequence input by iterating through the elements. RNNs pass the outputs from one timestep to their input—and then to the next.

The tf.keras.layers.Bidirectional wrapper can also be used with an RNN layer. This propagates the input forward and backwards through the RNN layer and then concatenates the output. This helps the RNN to learn long range dependencies.

model = tf.keras.Sequential([
    tf.keras.layers.Embedding(tokenizer.vocab_size, 64),
    tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(64)),
    tf.keras.layers.Dense(64, activation='relu'),
    tf.keras.layers.Dense(1, activation='sigmoid')
])

Compile the Keras model to configure the training process:

model.compile(loss='binary_crossentropy',
              optimizer='adam',
              metrics=['accuracy'])

Train the model

history = model.fit(train_dataset, epochs=10,
                    validation_data=test_dataset)
Epoch 1/10

W0614 18:51:00.588112 139970744243968 deprecation.py:323] From /home/kbuilder/.local/lib/python3.5/site-packages/tensorflow/python/ops/math_grad.py:1250: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.
Instructions for updating:
Use tf.where in 2.0, which has the same broadcast rule as np.where

391/391 [==============================] - 329s 842ms/step - loss: 0.5890 - accuracy: 0.6834 - val_loss: 0.0000e+00 - val_accuracy: 0.0000e+00
Epoch 2/10
391/391 [==============================] - 123s 315ms/step - loss: 0.4037 - accuracy: 0.8253 - val_loss: 0.4635 - val_accuracy: 0.7893
Epoch 3/10
391/391 [==============================] - 98s 251ms/step - loss: 0.3419 - accuracy: 0.8575 - val_loss: 0.4389 - val_accuracy: 0.8198
Epoch 4/10
391/391 [==============================] - 83s 211ms/step - loss: 0.2610 - accuracy: 0.9026 - val_loss: 0.4208 - val_accuracy: 0.8430
Epoch 5/10
391/391 [==============================] - 85s 217ms/step - loss: 0.2245 - accuracy: 0.9166 - val_loss: 0.4551 - val_accuracy: 0.8169
Epoch 6/10
391/391 [==============================] - 81s 207ms/step - loss: 0.1841 - accuracy: 0.9328 - val_loss: 0.5575 - val_accuracy: 0.8316
Epoch 7/10
391/391 [==============================] - 75s 192ms/step - loss: 0.1893 - accuracy: 0.9288 - val_loss: 0.4696 - val_accuracy: 0.8104
Epoch 8/10
391/391 [==============================] - 78s 199ms/step - loss: 0.2186 - accuracy: 0.9165 - val_loss: 0.5216 - val_accuracy: 0.8284
Epoch 9/10
391/391 [==============================] - 74s 189ms/step - loss: 0.1691 - accuracy: 0.9370 - val_loss: 0.5743 - val_accuracy: 0.8051
Epoch 10/10
391/391 [==============================] - 70s 178ms/step - loss: 0.3532 - accuracy: 0.8471 - val_loss: 0.5273 - val_accuracy: 0.7723
test_loss, test_acc = model.evaluate(test_dataset)

print('Test Loss: {}'.format(test_loss))
print('Test Accuracy: {}'.format(test_acc))
    391/Unknown - 19s 48ms/step - loss: 0.5273 - accuracy: 0.7723Test Loss: 0.5272858859328053
Test Accuracy: 0.7723199725151062

The above model does not mask the padding applied to the sequences. This can lead to skewness if we train on padded sequences and test on un-padded sequences. Ideally the model would learn to ignore the padding, but as you can see below it does have a small effect on the output.

If the prediction is >= 0.5, it is positive else it is negative.

def pad_to_size(vec, size):
  zeros = [0] * (size - len(vec))
  vec.extend(zeros)
  return vec
def sample_predict(sentence, pad):
  tokenized_sample_pred_text = tokenizer.encode(sample_pred_text)

  if pad:
    tokenized_sample_pred_text = pad_to_size(tokenized_sample_pred_text, 64)

  predictions = model.predict(tf.expand_dims(tokenized_sample_pred_text, 0))

  return (predictions)
# predict on a sample text without padding.

sample_pred_text = ('The movie was cool. The animation and the graphics '
                    'were out of this world. I would recommend this movie.')
predictions = sample_predict(sample_pred_text, pad=False)
print (predictions)
[[0.75800496]]
# predict on a sample text with padding

sample_pred_text = ('The movie was cool. The animation and the graphics '
                    'were out of this world. I would recommend this movie.')
predictions = sample_predict(sample_pred_text, pad=True)
print (predictions)
[[0.5070383]]
plot_graphs(history, 'accuracy')

png

plot_graphs(history, 'loss')

png

Stack two or more LSTM layers

Keras recurrent layers have two available modes that are controlled by the return_sequences constructor argument:

  • Return either the full sequences of successive outputs for each timestep (a 3D tensor of shape (batch_size, timesteps, output_features)).
  • Return only the last output for each input sequence (a 2D tensor of shape (batch_size, output_features)).
model = tf.keras.Sequential([
    tf.keras.layers.Embedding(tokenizer.vocab_size, 64),
    tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(
        64, return_sequences=True)),
    tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(32)),
    tf.keras.layers.Dense(64, activation='relu'),
    tf.keras.layers.Dense(1, activation='sigmoid')
])
model.compile(loss='binary_crossentropy',
              optimizer='adam',
              metrics=['accuracy'])
history = model.fit(train_dataset, epochs=10,
                    validation_data=test_dataset)
Epoch 1/10
391/391 [==============================] - 647s 2s/step - loss: 0.5588 - accuracy: 0.7058 - val_loss: 0.0000e+00 - val_accuracy: 0.0000e+00
Epoch 2/10
391/391 [==============================] - 236s 603ms/step - loss: 0.4772 - accuracy: 0.7784 - val_loss: 0.4350 - val_accuracy: 0.8188
Epoch 3/10
391/391 [==============================] - 184s 469ms/step - loss: 0.3539 - accuracy: 0.8483 - val_loss: 0.4786 - val_accuracy: 0.7805
Epoch 4/10
391/391 [==============================] - 157s 401ms/step - loss: 0.2864 - accuracy: 0.8888 - val_loss: 0.4091 - val_accuracy: 0.8321
Epoch 5/10
391/391 [==============================] - 159s 406ms/step - loss: 0.2057 - accuracy: 0.9245 - val_loss: 0.4870 - val_accuracy: 0.8420
Epoch 6/10
391/391 [==============================] - 145s 371ms/step - loss: 0.1620 - accuracy: 0.9447 - val_loss: 0.5111 - val_accuracy: 0.8297
Epoch 7/10
391/391 [==============================] - 136s 347ms/step - loss: 0.1506 - accuracy: 0.9495 - val_loss: 0.7015 - val_accuracy: 0.7487
Epoch 8/10
391/391 [==============================] - 132s 337ms/step - loss: 0.1227 - accuracy: 0.9612 - val_loss: 0.6219 - val_accuracy: 0.8019
Epoch 9/10
391/391 [==============================] - 123s 314ms/step - loss: 0.0971 - accuracy: 0.9702 - val_loss: 0.6324 - val_accuracy: 0.8044
Epoch 10/10
391/391 [==============================] - 131s 334ms/step - loss: 0.0818 - accuracy: 0.9760 - val_loss: 0.6319 - val_accuracy: 0.8107
test_loss, test_acc = model.evaluate(test_dataset)

print('Test Loss: {}'.format(test_loss))
print('Test Accuracy: {}'.format(test_acc))
    391/Unknown - 33s 85ms/step - loss: 0.6319 - accuracy: 0.8107Test Loss: 0.6318796596792348
Test Accuracy: 0.8107200264930725
# predict on a sample text without padding.

sample_pred_text = ('The movie was not good. The animation and the graphics '
                    'were terrible. I would not recommend this movie.')
predictions = sample_predict(sample_pred_text, pad=False)
print (predictions)
[[0.07144996]]
# predict on a sample text with padding

sample_pred_text = ('The movie was not good. The animation and the graphics '
                    'were terrible. I would not recommend this movie.')
predictions = sample_predict(sample_pred_text, pad=True)
print (predictions)
[[0.10525462]]
plot_graphs(history, 'accuracy')

png

plot_graphs(history, 'loss')

png

Check out other existing recurrent layers such as GRU layers.