Alert button

A Deep Ensemble Model with Slot Alignment for Sequence-to-Sequence Natural Language Generation

May 16, 2018
Juraj Juraska, Panagiotis Karagiannis, Kevin K. Bowden, Marilyn A. Walker

Figure 1 for A Deep Ensemble Model with Slot Alignment for Sequence-to-Sequence Natural Language Generation
Figure 2 for A Deep Ensemble Model with Slot Alignment for Sequence-to-Sequence Natural Language Generation
Figure 3 for A Deep Ensemble Model with Slot Alignment for Sequence-to-Sequence Natural Language Generation
Figure 4 for A Deep Ensemble Model with Slot Alignment for Sequence-to-Sequence Natural Language Generation

Share this with someone who'll enjoy it:

Natural language generation lies at the core of generative dialogue systems and conversational agents. We describe an ensemble neural language generator, and present several novel methods for data representation and augmentation that yield improved results in our model. We test the model on three datasets in the restaurant, TV and laptop domains, and report both objective and subjective evaluations of our best model. Using a range of automatic metrics, as well as human evaluators, we show that our approach achieves better results than state-of-the-art models on the same datasets.

* Accepted to NAACL 2018  
View paper onarxiv icon

Share this with someone who'll enjoy it: