NORMA eResearch @NCI Library

Neural Text-to-Text Generation System using Generative Adversarial Network and Monte Carlo Policy Gradient

Moorthy, Seemanthini Narasimha (2020) Neural Text-to-Text Generation System using Generative Adversarial Network and Monte Carlo Policy Gradient. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
PDF (Master of Science)
Download (1MB) | Preview
[thumbnail of Configuration manual]
PDF (Configuration manual)
Download (1MB) | Preview


Text-to-text generation is a fundamental task in natural language processing. Traditional models rely on standalone recurrant neural networks like Long Short Term Memory(LSTM) and Gated Recurrent Units(GRU). Generative Adversarial Networks (GAN) have found little success in generating discrete valued data like text. Major drawbacks lies in the failure to pass discrete output from generator model to discriminator model, and the inability of the discriminator model to assess incomplete sentences. This research strengthens the use of Generative Adversarial Networks combining it with Monte Carlo Policy Gradient, where the gradient policy update comes directly from the discriminator model and is passed back as the reward signal by using Monte Carlo Search algorithm. The results show that by combining Generative Adversarial Networks and Reinforce Algorithm, significant results can be obtained comparative to baseline models using evaluation metric called Bilingual Evaluation Understudy Score -N (BLEU-N). A BLEU score of 0.3 was achieved overall through different experiments.
Keywords: Natural Language Processing, Long Short Term Memory(LSTM), Gated Recurrent Units(GRU), Generative Adversarial Networks(GAN), Monte Carlo Policy Gradient

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Q Science > QA Mathematics > Computer software
T Technology > T Technology (General) > Information Technology > Computer software
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Dan English
Date Deposited: 17 Jun 2020 15:33
Last Modified: 17 Jun 2020 15:33

Actions (login required)

View Item View Item