This course is an introduction to Social Signal Processing, the field aimed at bridging the social intelligence gap between people and machines. At its core, social intelligence consists of sensing nonverbal behavioral cues displayed by people around us (facial expressions, gestures, vocalizations, postures), interpreting these cues in terms of social signals (relational attitudes like interest, hostility, empathy, agreement and disagreement, or dominance), and displaying as a response natural, consistent behaviors (interest for those we are interested in, or hostility for those we are hostile to).

From a scientific point of view, this results into three core objectives:

  1. Detection of nonverbal behavioral cues using sensors including microphones, cameras, proximity detectors, or others.
  2. Inference of social signals from nonverbal behavioral cues.
  3. Synthesis of social signals through different forms of embodiment in artificial agents, avatars, or synthetic voices.

    The course focuses on the first two questions and provides an introduction to the main scientific and technological problems and the existing work for two major scenarios, namely face-to-face interactions (meetings, conversations, etc.) and large scale social interactions (daily life of populations sensed with mobile devices).

    This interdisciplinary research has the potential of significantly advancing several domains related to automatic monitoring, including video surveillance, ambient intelligence, marketing, office space design, and architecture and urbanism.


    %d bloggers like this: