You are here

Princeton University Machine Learning for the Sciences

Solving Inverse Problems with Data-driven Priors

I will present a Bayesian machine learning architecture that combines a physically motivated parameterization and an analytic error model for the likelihood with a deep generative model providing a powerful data-driven prior for complex signals. This combination yields an interpretable and differentiable generative model, allows the incorporation of prior knowledge, and can be utilized for observations with different data quality without having to retrain the deep network. I will demonstrate this approach with an example of astronomical source separation in current imaging data, yielding a physical and interpretable model of astronomical scenes.

Featuring

Peter Melchior

Speaker Affiliation

Princeton University

Notes

Lunch will be served.

Additional Information

Date & Time
January 17, 2020 | 12:001:00pm

Location

Center for Statistics and Machine Learning (CSML) 26 Prospect Ave Auditorium 103