Room 333 Wisenbaker (fishbowl)
Dr. Lu Su, (SUNY Buffalo)
Abstract: The recent proliferation of increasingly capable and affordable mobile devices has given rise to crowd sensing, a newly emerged sensing paradigm where the collection of sensory data is outsourced to a crowd of participants who report what they learned about the physical and social world. Despite their tremendous benefits, crowd sensing systems pose a great new research challenge. That is, human-carried sensors are not reliable, due to various possible reasons including incomplete observations, background noise, poor sensor quality, or even the intent to deceive. To address this sensor reliability problem, we can integrate the data from multiple sensors, as this will likely cancel out the errors of individual sensors. However, the straightforward data integration methods, such as voting/averaging, treat all the sensors equally, and fail to capture the variety in the Quality of Information (QoI) of different sensors. In this talk, I will introduce a novel quality aware big data integration framework that can jointly estimate the QoI of each sensor and aggregate the big sensory data weighted by the estimated QoI. The proposed framework addresses the emerging challenge brought by the unreliable and noisy crowdsourced data, and thus can benefit a whole spectrum of crowd sensing applications, such as smart transportation, environment monitoring, urban sensing, health care, and many others.
Bio: Dr. Lu Su is an assistant professor in the department of Computer Science and Engineering at the State University of New York at Buffalo (SUNY Buffalo). He obtained Ph.D. in Computer Science, and M.S. in Statistics, both from the University of Illinois at Urbana-Champaign, in 2013 and 2012, respectively. He has also worked at IBM T. J. Watson Research Center and National Center for Supercomputing Applications. Dr. Su’s research interests lie in the general areas of cyber-physical systems, internet of things, crowd and social sensing systems, mobile computing, wireless and sensor networks, security and privacy, data mining and machine learning, and renewable energy. With the ultimate goal of building information-effective and resource-efficient systems that interconnect human beings and the surrounding physical world, he focuses his current research on developing theories, algorithms, and tools that can intelligently collect, transmit, integrate, and eventually transform the deluge of sensory data generated by the ubiquitous human and physical sensors into high quality information that can draw a better understanding of the social and physical world. Please refer to his homepage for more information: http://www.cse.buffalo.edu/~lusu/.