BEGIN:VCALENDAR VERSION:2.0 PRODID:researchseminars.org CALSCALE:GREGORIAN X-WR-CALNAME:researchseminars.org BEGIN:VEVENT SUMMARY:Anandaroop Ray (Geoscience Australia) DTSTART:20210628T053000Z DTEND:20210628T070000Z DTSTAMP:20260404T095120Z UID:RSES_Geophysics/1 DESCRIPTION:Title: Bayesian inference using trans-D Gaussian processes using t rans-D Gaussian processes\nby Anandaroop Ray (Geoscience Australia) as part of ANU Research School of Earth Sciences Geophysics Seminars\n\nLect ure held in Jaeger 1 Seminar Room.\n\nAbstract\nTo understand earth proces ses\, geoscientists infer subsurface earth properties such as electromagne tic resistivity or seismic velocity from surface observations such as magn etotelluric data or seismograms. These properties are used to populate an earth model vector\, and the spatial variation of properties sheds light o n underlying earth structure and associated geodynamic phenomena. I will s how that making accurate inferences about earth properties can require the use of a multilevel parameterisation\, where the same algorithm is used i n a nested fashion to infer all levels of the parameterisation. This is an “infer to infer” paradigm analogous to the “learning to learn” pa radigm commonplace in the machine learning literature. A non-stationary tr ans-dimensional Gaussian Process (TDGP) is used to parameterise earth prop erties (e.g.\, resistivity)\, and a multi-channel stationary TDGP is used to parameterise the length scales of the top-level parameterisation. Using non-stationary GP kernels and efficient approximations to update GP kerne l matrices\, sharp discontinuities as well as smooth regions can be repres ented within one earth model. As GPs are multi-dimensional interpolators\, the same theory and computer code can be used to solve geophysical proble ms in 1D\, 2D and 3D.\n LOCATION:https://stable.researchseminars.org/talk/RSES_Geophysics/1/ END:VEVENT END:VCALENDAR