Natural Language Models For Unnatural Languages

Event Start
Event End
B1, L2, R2202


Generative models for language generation, particularly based on transformers have shown remarkable performance in domains dealing with language.

This dissertation aims to describe our framework that adapts generative models of languages, allowing them to generate non-language objects. This proposed framework allows for generation in a common framework across seemingly unrelated domains. The dissertation tackles three problems of increasing complexity - room-layout generation, CAD-sketch generation, and furniture-layout generation. We solve each of the problems by creating a grammar that describes each object in the problem domain representing each object by a sequence of tokens.

Brief Biography

Wamiq Para is an MS/Ph.D. student in the Computer Science Program under the supervision of Professor Peter Wonka at the Visual Computing Center (VCC) at King Abdullah University of Science and Technology (KAUST). Wamiq earned his Bachelor's Degree in Electronics and Communications from the National Institute of Technology, Srinagar. He is interested in Sequence Generation, Image Editing and Generative Modeling in general.

Contact Person