Louis-Philippe Morency
- Published in print:
- 2013
- Published Online:
- January 2014
- ISBN:
- 9780195387643
- eISBN:
- 9780199369195
- Item type:
- chapter
- Publisher:
- Oxford University Press
- DOI:
- 10.1093/acprof:oso/9780195387643.003.0009
- Subject:
- Psychology, Cognitive Models and Architectures, Cognitive Psychology
Face-to-face communication is highly interactive. Even when only one person speaks at the time, other participants exchange information continuously amongst themselves and with the speaker through ...
More
Face-to-face communication is highly interactive. Even when only one person speaks at the time, other participants exchange information continuously amongst themselves and with the speaker through gesture, gaze, posture and facial expressions. This chapter argues that it is possible to significantly improve state-of-the art recognition techniques by exploiting regularities in how people communicate. For example, listeners are far more likely to nod or shake if the speaker has just asked them a question, and incorporating such dialogue context can improve recognition performance during human-robot interaction. The chapter introduces the idea of encoding dictionary, a technique for contextual feature representation inspired by the influence speaker context has on the listener feedback. Automatic selection of relevant contextual features is performed by looking at individual and joint influences of context. The final contextual integration is done using a discriminative sequential model. The chapter shows the importance of context in affective behavior understanding on two different domains: interaction with a robot and human dyadic interaction.Less
Face-to-face communication is highly interactive. Even when only one person speaks at the time, other participants exchange information continuously amongst themselves and with the speaker through gesture, gaze, posture and facial expressions. This chapter argues that it is possible to significantly improve state-of-the art recognition techniques by exploiting regularities in how people communicate. For example, listeners are far more likely to nod or shake if the speaker has just asked them a question, and incorporating such dialogue context can improve recognition performance during human-robot interaction. The chapter introduces the idea of encoding dictionary, a technique for contextual feature representation inspired by the influence speaker context has on the listener feedback. Automatic selection of relevant contextual features is performed by looking at individual and joint influences of context. The final contextual integration is done using a discriminative sequential model. The chapter shows the importance of context in affective behavior understanding on two different domains: interaction with a robot and human dyadic interaction.
