Thomas Hermans, Ghent University-Bayesian Evidential Learning: Get Away with Geophysical Inversion
- Thursday, Nov 5, 2020 12:00 PM
- https://stanford.zoom.us/s/95742837634 Passcode: 840912
- Geophysics Department
Department of Geology, Ghent University, Belgium
Bayesian Evidential Learning or How to (Almost) Get Away with Geophysical Inversion Issues
Imaging the subsurface of the Earth is the main objective of geophysicists. Deterministic inversions lack the ability to produce satisfactory quantification of uncertainty, whereas stochastic inversions are computationally (too) demanding. In this contribution, a new efficient stochastic method to interpret geophysical data avoiding direct inversion through machine learning is proposed. The methodology based on Bayesian Evidential Learning will be illustrated on different 1D, 2D and 3D examples from different data types (SNMR, ERT, surface wave seismic).
https://stanford.zoom.us/s/95742837634 Passcode: 840912