Co-sponsored by: Ms. Salwa

Fuelled by the sheer volumes of data and computing power, the rapidly progressing developments in artificial intelligence and machine learning led to recognition algorithms with unprecedented performances. Current image recognition performances are nearing a level approaching that of human performance. On some specialised domains, computers even outperform humans in image recognition abilities. The impressive developments enable a next step in the integration of computers and computing devices in our everyday world. In order to achieve a seamless integration of computers in our daily life, computers should be able to communicate in a human-like manner. The long-standing challenge to achieve human-level verbal communication, i.e., speech recognition and speech production, obscured the importance of its non-verbal counterpart. Non-verbal communication is crucial to the social interactions between humans. Examples of nonverbal expressions are vocal expression, such as vocal pitch or intensity, and facial expressions such as smiling or frowning. Nonverbal expressions provide indispensable contextual cues to social interactions. Our research focusses on the understanding, modelling and simulation of the interactive nonverbal dynamics of communicating humans, with the aim to develop computers, robots or intelligent agents with social capabilities. We formalise the nonverbal interactions between human-human or human-computer dyads in terms of dynamical systems theory. In this formalisation, dyadic nonverbal communication gives rise to an attractor manifold representing the complex expression dynamics of both interlocutors. This allows us to study and simulate the causal interactions between communicating faces and voices by means of the powerful tools offert by dynamical systems theory. The result is a characterisation of the dynamical building blocks of nonverbal communication. In the presentation the results obtained will be demonstrated and their contribution to the development of social computers will be explained by means of illustrative examples. 

Speaker(s): Professor Eric Postma,

Location:
Room: Hall A – Main Building
Bldg: American University of Sharjah
Sharjah, United Arab Emirates