Details
| Original language | English |
|---|---|
| Pages (from-to) | 393-412 |
| Number of pages | 20 |
| Journal | Proceedings of Machine Learning Research |
| Volume | 121 |
| Publication status | Published - 2020 |
| Event | 3rd Conference on Medical Imaging with Deep Learning, MIDL 2020 - Virtual, Online, Canada Duration: 6 Jul 2020 → 8 Jul 2020 |
Abstract
The consideration of predictive uncertainty in medical imaging with deep learning is of utmost importance. We apply estimation of predictive uncertainty by variational Bayesian inference with Monte Carlo dropout to regression tasks and show why predictive uncertainty is systematically underestimated. We suggest using σ scaling with a single scalar value; a simple, yet effective calibration method for both aleatoric and epistemic uncertainty. The performance of our approach is evaluated on a variety of common medical regression data sets using different state-of-the-art convolutional network architectures. In all experiments, σ scaling is able to reliably recalibrate predictive uncertainty. It is easy to implement and maintains the accuracy. Well-calibrated uncertainty in regression allows robust rejection of unreliable predictions or detection of out-of-distribution samples. Our source code is available at: github.com/mlaves/well-calibrated-regression-uncertainty.
Keywords
- Bayesian approximation, variational inference
ASJC Scopus subject areas
- Computer Science(all)
- Artificial Intelligence
- Computer Science(all)
- Software
- Engineering(all)
- Control and Systems Engineering
- Mathematics(all)
- Statistics and Probability
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
In: Proceedings of Machine Learning Research, Vol. 121, 2020, p. 393-412.
Research output: Contribution to journal › Conference article › Research › peer review
}
TY - JOUR
T1 - Well-Calibrated Regression Uncertainty in Medical Imaging with Deep Learning
AU - Laves, Max Heinrich
AU - Ihler, Sontje
AU - Fast, Jacob F.
AU - Kahrs, Lüder A.
AU - Ortmaier, Tobias
N1 - Funding Information: We thank Vincent Modes for his insightful comments. This research has received funding from the European Union as being part of the ERDF OPhonLas project.
PY - 2020
Y1 - 2020
N2 - The consideration of predictive uncertainty in medical imaging with deep learning is of utmost importance. We apply estimation of predictive uncertainty by variational Bayesian inference with Monte Carlo dropout to regression tasks and show why predictive uncertainty is systematically underestimated. We suggest using σ scaling with a single scalar value; a simple, yet effective calibration method for both aleatoric and epistemic uncertainty. The performance of our approach is evaluated on a variety of common medical regression data sets using different state-of-the-art convolutional network architectures. In all experiments, σ scaling is able to reliably recalibrate predictive uncertainty. It is easy to implement and maintains the accuracy. Well-calibrated uncertainty in regression allows robust rejection of unreliable predictions or detection of out-of-distribution samples. Our source code is available at: github.com/mlaves/well-calibrated-regression-uncertainty.
AB - The consideration of predictive uncertainty in medical imaging with deep learning is of utmost importance. We apply estimation of predictive uncertainty by variational Bayesian inference with Monte Carlo dropout to regression tasks and show why predictive uncertainty is systematically underestimated. We suggest using σ scaling with a single scalar value; a simple, yet effective calibration method for both aleatoric and epistemic uncertainty. The performance of our approach is evaluated on a variety of common medical regression data sets using different state-of-the-art convolutional network architectures. In all experiments, σ scaling is able to reliably recalibrate predictive uncertainty. It is easy to implement and maintains the accuracy. Well-calibrated uncertainty in regression allows robust rejection of unreliable predictions or detection of out-of-distribution samples. Our source code is available at: github.com/mlaves/well-calibrated-regression-uncertainty.
KW - Bayesian approximation, variational inference
UR - http://www.scopus.com/inward/record.url?scp=85093069704&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85093069704
VL - 121
SP - 393
EP - 412
JO - Proceedings of Machine Learning Research
JF - Proceedings of Machine Learning Research
T2 - 3rd Conference on Medical Imaging with Deep Learning, MIDL 2020
Y2 - 6 July 2020 through 8 July 2020
ER -