Learning to Generate Compositional Color Descriptions

Will Monroe, Noah D. Goodman, Christopher Potts
2016 Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing  
The production of color language is essential for grounded language generation. Color descriptions have many challenging properties: they can be vague, compositionally complex, and denotationally rich. We present an effective approach to generating color descriptions using recurrent neural networks and a Fouriertransformed color representation. Our model outperforms previous work on a conditional language modeling task over a large corpus of naturalistic color descriptions. In addition, probing
more » ... the model's output reveals that it can accurately produce not only basic color terms but also descriptors with non-convex denotations ("greenish"), bare modifiers ("bright", "dull"), and compositional phrases ("faded teal") not seen in training.
doi:10.18653/v1/d16-1243 dblp:conf/emnlp/MonroeGP16 fatcat:nlgku4tlrbgbpp7dwz3zq74twy