dem-esgal / Text-Classification

Implementation of papers for text classification task on DBpedia

Text-Classification

Implement some text classification model with TensorFlow based on some papers:

Attention-Based Bidirection LSTM for Text Classification

Paper: Attention-Based Bidirectional Long Short-Term Memory Networks for Relation Classification

See ABBLSTM.py

Hierarchical Attention Networks for Text Classification

Paper: Hierarchical Attention Networks for Document Classification

See ALSTM_Hierarchical.py

Adversarial Training Methods For Supervised Text Classification

Paper: Adversarial Training Methods For Semi-Supervised Text Classification

See adversarial_abblstm.py

Dataset and Performance

You can load the data with

 dbpedia = tf.contrib.learn.datasets.load_dataset('dbpedia', test_with_fake_data=FLAGS.test_with_fake_data)

The ABBLSTM model can get a 98.23 % accuracy on the dataset.

And the Aderversarial ABBLSTM achieves 98.5% accuracy, training with AWS p2 instance for almost 2 hours.

Reference

Learn a lot from ilivans/tf-rnn-attention .