Andreas Krause, Assistant Professor of Computer Science, is an invited speaker at the National Academy of Sciences' Kavli Frontiers of Science Symposium. This annual symposium brings together some of the very best young scientists to discuss exciting advances and opportunities in their fields. Speakers are urged to focus their talks on current cutting-edge research in their disciplines to colleagues outside their field. Professor Krause's presentation is entitle Optimizing Sensing for Decision Making and he will discuss how harnessing sensing resources, such as cell phones and navigations devices, could have enormous benefit on the productivity, quality and security of our society.
Presentation Title: Optimizing Sensing for Decision Making Presentation Abstract: Sensors are everywhere: Examples include community-held sensors such as accelerometers in cell phones, GPS receivers and navigation devices in cars, infrastructural sensors such as smart meters in the power grid, sensors for environmental monitoring and many others. Harnessing these sensing resources could have enormous benefit on the productivity, quality and security of our society. In order to make use of these resources, we need to address important research challenges: How can we model and robustly reason about data obtained from heterogeneous, noisy sensors? How can we efficiently make informed, distributed decisions under uncertainty? How can we cope with constraints due to limited battery, computational power and communication capability? How can we extract most useful information from the massive amounts of data originating from large-scale sensor and information networks? In this talk, I will discuss some of these challenges and possible approaches to address them, based on statistical inference, discrete optimization and Bayesian experimental design. I will illustrate them in the context of real-world sensing problems, including autonomous environmental monitoring, protecting drinking water distribution networks, and earthquake detection from community-held accelerometers.