TY - GEN
T1 - Solving Inverse Problems with a Flow-based Noise Model
AU - Whang, Jay
AU - Lei, Qi
AU - Dimakis, Alexandros G.
N1 - Publisher Copyright:
Copyright © 2021 by the author(s)
PY - 2021
Y1 - 2021
N2 - We study image inverse problems with a normalizing flow prior. Our formulation views the solution as the maximum a posteriori estimate of the image conditioned on the measurements. This formulation allows us to use noise models with arbitrary dependencies as well as non-linear forward operators. We empirically validate the efficacy of our method on various inverse problems, including compressed sensing with quantized measurements and denoising with highly structured noise patterns. We also present initial theoretical recovery guarantees for solving inverse problems with a flow prior.
AB - We study image inverse problems with a normalizing flow prior. Our formulation views the solution as the maximum a posteriori estimate of the image conditioned on the measurements. This formulation allows us to use noise models with arbitrary dependencies as well as non-linear forward operators. We empirically validate the efficacy of our method on various inverse problems, including compressed sensing with quantized measurements and denoising with highly structured noise patterns. We also present initial theoretical recovery guarantees for solving inverse problems with a flow prior.
UR - http://www.scopus.com/inward/record.url?scp=85161309640&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85161309640&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85161309640
T3 - Proceedings of Machine Learning Research
SP - 11146
EP - 11157
BT - Proceedings of the 38th International Conference on Machine Learning, ICML 2021
PB - ML Research Press
T2 - 38th International Conference on Machine Learning, ICML 2021
Y2 - 18 July 2021 through 24 July 2021
ER -