Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox


Unsupervised Natural Language Generation with Denoising Autoencoders

Aug 24, 2018
Markus Freitag, Scott Roy


Share this with someone who'll enjoy it:


Generating text from structured data is important for various tasks such as question answering and dialog systems. We show that in at least one domain, without any supervision and only based on unlabeled text, we are able to build a Natural Language Generation (NLG) system with higher performance than supervised approaches. In our approach, we interpret the structured data as a corrupt representation of the desired output and use a denoising auto-encoder to reconstruct the sentence. We show how to introduce noise into training examples that do not contain structured data, and that the resulting denoising auto-encoder generalizes to generate correct sentences when given structured data.

* Accepted at EMNLP 2018 


   Access Paper Source



Share this with someone who'll enjoy it: