(Nov 27) Human-Centric Sensing and Analytics: Cases, Challenges, and Opportunities

Human-centric sensing applications have rapidly penetrated every facet of our daily lives including child education, elderly support, smart transportation, and automated homes. They have evolved to capture and use a rich set of human activities and internal states (such as intention, engagement, attention, depression) to provide highly usable situational services. For example, an upcoming mobile advertisement application will precisely target customers with the buying intention and send promotions exactly when they are free to look at the coupons. Over the years, I have been working on core challenges to (1) innovate mobile/wearable/IoT systems to enable highly enriched, and available context awareness, (2) design innovative human-centric sensing applications, and (3) deploy the systems and applications to real target users at scale and make them highly usable. In this talk, I will first share key challenges and opportunities in human-centric sensing and analytics with example systems and applications that I have built and deployed. Then, I will introduce two specific systems I have been working on: (1) DeepMon, the mobile deep learning system to enable continuous vision sensing application, and (2) CommBetter, the real-time face-to-face interaction sensing system.