Work

Learning to Generate Compositional Color Descriptions

Year: 2016
Type: conference-paper
Abstract: 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 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.
Institution: Stanford University
Language: English
Field: Psychology
Open Access status: hybrid
OpenAlex ID: W2429192533
FWCI: 10.7
Cites: 20
Cited by: 24
Related to: 10