Boosting Naturalness of Language in Task-oriented Dialogues via Adversarial Training
arXiv:2004.14565
Abstract
The natural language generation (NLG) module in a task-oriented dialogue system produces user-facing utterances conveying required information. Thus, it is critical for the generated response to be natural and fluent. We propose to integrate adversarial training to produce more human-like responses. The model uses Straight-Through Gumbel-Softmax estimator for gradient computation. We also propose a two-stage training scheme to boost performance. Empirical results show that the adversarial training can effectively improve the quality of language generation in both automatic and human evaluations. For example, in the RNN-LG Restaurant dataset, our model AdvNLG outperforms the previous state-of-the-art result by 3.6% in BLEU.
arXiv commentSIGDial 2020
References in corpus (13)
- Sequence to Sequence Learning with Neural Networks
- An overview of gradient descent optimization algorithms
- Learning Phrase Representations using RNN Encoder-Decoder for Statistical Machine Translation
- Categorical Reparameterization with Gumbel-Softmax
- Sequence Level Training with Recurrent Neural Networks
- Deep Reinforcement Learning for Dialogue Generation
- Adversarial Learning for Neural Dialogue Generation
- Semantically Conditioned LSTM-based Natural Language Generation for Spoken Dialogue Systems
- Adversarial Generation of Natural Language
- Adversarial Neural Machine Translation
- Unsupervised Text Style Transfer using Language Models as Discriminators
- Multi-domain Neural Network Language Generation for Spoken Dialogue Systems
- ARAML: A Stable Adversarial Training Framework for Text Generation