The Beckman Institute Fellows program allows recent Ph.D.s an opportunity to do highly focused, interdisciplinary research without the burdens of teaching or taking classes. The four 2005 Beckman Fellows are about midway through their appointments— long enough to have some perspective on their experience but at a stage where a post-Fellows life is growing closer. Beckman Institute writer Steve McGaughey posed questions about their present work and future plans to the quartet of 2005 Beckman Fellows: Chandramallika Basak, Silvio Savarese, Emma Falck, and Zhihong Zeng.
Silvio Savarese
Silvio is a member of the Human-Computer Intelligent Interaction research initiative who earned his Ph.D. at the California Institute of Technology. He researches 3-D scene modeling and understanding, human visual perception, and other aspects of how the interactions of computers and human beings can be improved.
Q: Has your research gone as expected when you first applied for the appointment, or has it changed in some ways?
A: In a way my research has gone as expected. When I started I was very excited by the idea of collaborating with other faculty and to try different research topics, and this is what happened. I have been working on different parallel projects, and my first year was both exciting and challenging. It was setting the basis for all my future research and future goals. Now I think that things are finally coming to the point of giving us interesting results and in the next few months we'll be able to present the results. The nice thing is being able to carry out independent research according to my own agenda thanks to the Fellows appointment. I don't have to teach and can carry out my goals and find the right directions and right collaborations.
Q: Where do you see yourself in 5 years?
A: I would like to pursue my career in academia as a faculty member at a nice school (laughs), hopefully. My research area is computer vision and I hope that in five years I can bring some significant contributions to this field and try to accomplish what people in it are trying to do, namely to make computers see and understand words. Hopefully we will be close to that in a few years.
Zhihong Zeng
After earning a Ph.D. from the Institute of Automation, Chinese Academy of Sciences, Zhihong began working in the Human- Computer Intelligent Interaction research initiative at Beckman. He studies multimodal emotion assessment in naturally occurring settings toward advancing the human-computer interaction experience.
Q: Has your research gone as expected when you first applied for the appointment, or has it changed in some ways?
A: In some ways I have to say that my current research is a little different from what I expected because I have a better understanding of my research problem, and found it was more difficult than what I expected. I think that I am lucky to work at the Beckman Institute because it enables me to have immediate access to expertise in the related areas. I always get encouragement from the faculty here (Thomas Huang, Steven Levinson, and Dan Roth) and from Glen Roisman from the Psychology Department. That is the reason why I can still make progress even though it is a challenging problem.
Q: Where do you see yourself in 5 years?
A: I hope that I will still be here after the Fellows appointment ends because I like working here and enjoy Beckman's research environment atmosphere. The other aspect is where will I be in the research field. Automatic emotion recognition is a challenging and largely unexplored problem. I believe that in five years we will have a better understanding of human emotion expression and perception, and the advances in techniques will enable us to apply this understanding in automatic emotion recognition. At that time I expect that most of the current work, which focuses on recognizing pre-segmented, posed, and basic emotions by using a multimodal approach, should allow us to build a multimodal, continuous, spontaneous emotion recognition system. This system will contribute to a new paradigm for human computer interaction, and make emotion analysis more efficient and objective, thereby contributing to some related areas of research.