Event time:
Wednesday, June 20, 2018 - 2:00pm
Location:
LOM 215
Speaker:
Hau-tieng Wu
Speaker affiliation:
Duke University
Event description:
Quantifying the intrinsic structure from a given massive dataset is a commonly encountered challenge in medical society. The problem becomes more challenging due to the advanced sensing techniques; for example, multiple sensors with heterogenous data types. Diffusion geometry is a flexible framework for this kind of challenge that has led to several convincing results. Some topics in diffusion geometry, including diffusion map, alternating diffusion and its theoretical background will be discussed. Its application to explore sleep geometry will be discussed.