Dropout as a Bayesian approximation: Representing model uncertainty in deep learning
- Submitting institution
-
University of Oxford
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 2040
- Type
- E - Conference contribution
- DOI
-
-
- Title of conference / published proceedings
- 33rd International Conference on Machine Learning, ICML 2016
- First page
- 1651
- Volume
- 3
- Issue
- -
- ISSN
- 2640-3498
- Open access status
- Compliant
- Month of publication
- January
- Year of publication
- 2016
- URL
-
-
- 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
-
1
- Research group(s)
-
-
- Citation count
- -
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- The tools we developed in this paper are wide reaching, being used both within academia and in industry. Within academia, our tools have been applied to the medical domain: for example, in nodule segmentation in Multiple Sclerosis MRI scans (Arbel et al. In MICCAI’18.), and for diabetic retinopathy diagnosis in a work published in Nature (Leibig et al., Nature, 2017). In astrophysics our work is used in parameter estimation for strong gravitational lensing (Levasseur et al. Astrophysical Journal Letters, 2017). It is also used by Toyota in the development of self-driving cars (Rus et al., BDL workshop at NIPS’17).
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -