Variational inference for computational imaging inverse problems
- Submitting institution
-
University of Glasgow
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11-12033
- Type
- D - Journal article
- DOI
-
-
- Title of journal
- Journal of Machine Learning Research
- Article number
- -
- First page
- 1
- Volume
- 21
- Issue
- 179
- ISSN
- 1532-4435
- Open access status
- Compliant
- Month of publication
- September
- Year of publication
- 2020
- URL
-
http://eprints.gla.ac.uk/223297/
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
-
- Citation count
- 0
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- ORIGINALITY: Proposes a novel way to train Bayesian variational inference for inverse problems exploiting uncertain domain knowledge/models in combination with smaller sets of experimentally collected data. RIGOUR: Theoretical development, implementation and testing on multiple real-world and simulated benchmarks. SIGNIFICANCE: renders Bayesian machine learning (giving not just means, but uncertainty via full posterior distribution) accessible for a broad range of imaging applications, where empirical training data is typically scarce or expensive to collect, to solve real-world imaging inverse problems with minimal data collection efforts. Published in the top ML journal, it redefines state-of-the-art for important physics problems.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
