U-M graduate student Yuwei "Emily" Bao (Computer Science and Engineering) will present "Learning to Mediate Disparities Towards Pragmatic Communication."

ABSTRACT
Human communication is a collaborative process. Speakers, on top of conveying their own intent, adjust the content and language expressions by taking the listeners into account, including their knowledge background, personality, and physical capabilities. Towards building AI agents that have similar abilities in language communication, we propose a novel rational reasoning framework, Pragmatic Rational Speaker (PRS), where the speaker learns the speaker-listener disparity and adjusts the speech accordingly, by adding a light-weighted disparity adjustment layer into working memory on top of speaker’s long-term memory system. By fixing the long-term memory, the PRS only needs to update its working memory to learn and adapt to different types of listeners. To validate our framework, we create a dataset that simulates different types of speaker-listener disparities in the context of referential games. Our empirical results demonstrate that the PRS is able to shift its output towards the language that listeners are able to understand, significantly improve the collaborative task outcome, and learn the disparity faster than joint training.