Gualtiero Volpe, Casa Paganini – InfoMus Research Centre, DIBRIS , University of Genoa - Multimodal Systems for Embodied Experience of Music and Audiovisual Content
Abstract: Embodied cooperation “arises when two co-present individuals in motion coordinate their goal-directed actions”. The adoption of the embodied cooperation paradigm for the development of embodied and social multimedia systems opens new perspectives for social applications in future User Centric Media. At the same time, mobile technologies enable to share experiences in an easier and faster way, using multiple modalities and media. By combining these aspects, we propose mobile systems integrating real-time analysis of non-verbal expressive gesture and social behaviour. This keynote presents an overview of recent research results at Casa Paganini-InfoMus in these directions, including the EyesWeb XMI software platform. The focus is on socio-mobile embodied experience and retrieval of music and audiovisual content, on applications for education, and for rehabilitation, e.g., on teaching autistic children to recognise and express emotions by non-verbal full-body movement and gesture. The research is partially funded by the European Projects ILHAIRE, MIROR, ASC INCLUSION (7 Framework Programme ICT), and MetaBody (EU Culture).
Pushmeet Kohli, Microsoft Research Cambridge - Learning to Interact (Naturally) with (All) Users
Abstract: The last few decades have seen a dramatic increase in our use of computational devices/systems in our every day lives. We use systems like search engines for our information needs and gaming consoles for entertainments purposes. Our growing reliance on these technologies has motivated researchers to work on the development of "natural" user interfaces that can understand the intent and preferences of users and are easier to use. Making machines understand human intentions and preferences is an exceptionally challenging problem for the AI community. Part of this difficulty lies in capturing the large variability in preferences and behaviour of different users. In this talk, I will discuss some steps we have taken to overcome this problem in the context of two application scenarios: interaction/activity recognition using the Kinect, and information retrieval. I will start by discussing our work on human pose estimation using the Kinect sensor and discuss the challenges of developing a system that is supposed to work on "everybody". I will then discuss the problem of personalization in the context of information retrieval and discuss how traits like personality of users can be inferred from their online behaviour.
Gualtiero Volpe, PhD in Computer Engineering, teaches Multimodal Systems for Human-Computer Interaction and Computer Engineering Foundations. His research interests include multimodal interfaces, sound and music computing, computational models of non-verbal expressive gesture, emotion, and social signals, interactive multimodal systems for performing arts and cultural heritage, education, therapy and rehabilitation. He is President of AIMI (Italian Association for Musical Informatics). He was Guest Editor of special issues for Journal of New Music Research, Journal on Multimodal User Interfaces, and Entertainment Computing. He was co-chair of the 5th Intl Gesture Workshop, of the Intl Conference on New Interfaces for Musical Expression in 2008, of the eNTERFACE’09 summer workshop on multimodal interfaces, and of the Social Behaviour in Music Workshops. Author of over 100 international scientific publications, he is the scientific responsible of the MIROR and ILHAIRE EU-ICT projects for University of Genoa.Pushmeet Kohli is a research scientist in the Machine Learning and Perception group at Microsoft Research Cambridge, and an associate of the Psychometric Centre and Trinity Hall, University of Cambridge. Pushmeet’s research revolves around Intelligent Systems and Computational Sciences, and he publishes in the fields of Machine Learning, Computer Vision, Information Retrieval, and Game Theory. His current research interests include “human behaviour analysis” and the “prediction of user preferences”. Pushmeet is interested in designing autonomous and intelligent computer vision, bargaining and trading systems which learn by observing and interacting with users on social media sites such as Facebook. He is also investigating the use of new sensors such as KINECT for the problems of human pose estimation, scene understanding and robotics. Pushmeet has won a number of awards and prizes for his research. His PhD thesis, titled "Minimizing Dynamic and Higher Order Energy Functions using Graph Cuts", was the winner of the British Machine Vision Association’s “Sullivan Doctoral Thesis Award”, and was a runner-up for the British Computer Society's “Distinguished Dissertation Award”. Pushmeet’s was also one of the two United Kingdom nominees for the ERCIM Cor Baayen award in 2010. Pushmeet’s papers have appeared in SIGGRAPH, NIPS, ICCV, AAAI, CVPR, PAMI, IJCV, CVIU, ICML, AISTATS, AAMAS, UAI, ECCV, and ICVGIP and have won best paper awards in ICVGIP 2006, 2010 and ECCV 2010. His research has also been the subject of a number of articles in popular media outlets such as Forbes, The Economic Times, New Scientist and MIT Technology Review.