Tid: Onsdag 4 oktober 2017, kl. 15:00-17:00
Plats: C389, Södra huset, Frescati

Postseminarium följer direkt efter seminariet i institutionens pentry.

Abstract

One of the most fundamental aspects of spoken dialogue is the organization of speaking between the participants. Since it is difficult to speak and listen at the same time, interlocutors need to take turns speaking, and this turn-taking has to be coordinated somehow. This coordination is achieved using verbal and non-verbal signals, expressed in the face and voice, including syntax, prosody and gaze. Contrary to this, spoken dialogue systems typically use a very simplistic, silence-based model of turn-taking, which often results in interruptions or sluggish responses. In this talk, I will give an overview of several studies on how to model turn-taking in spoken interaction, with a special focus on multi-modal, human-robot interaction. These studies show that humans in interaction with a human-like robot make use of the same coordination signals typically found in studies on human-human interaction, and that it is possible to automatically detect and combine these cues to facilitate real-time coordination. The studies also show that humans react naturally to such signals when used by a robot, without being given any special instructions. Finally, I will present recent work on how Recurrent Neural Networks can be used to train a predictive, continuous model of turn-taking from human-human interaction data.

Information om Gabriel Skantze vid KTH

Hjärtligt välkomna!

Hatice Zora