Well-calibrated Model Uncertainty with Temperature Scaling for Dropout Variational Inference

Add code
Oct 01, 2019
Max-Heinrich Laves, Sontje Ihler, Karl-Philipp Kortmann, Tobias Ortmaier
Figure 1 for Well-calibrated Model Uncertainty with Temperature Scaling for Dropout Variational Inference
Figure 2 for Well-calibrated Model Uncertainty with Temperature Scaling for Dropout Variational Inference
Figure 3 for Well-calibrated Model Uncertainty with Temperature Scaling for Dropout Variational Inference
Figure 4 for Well-calibrated Model Uncertainty with Temperature Scaling for Dropout Variational Inference

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon

In this paper, well-calibrated model uncertainty is obtained by using temperature scaling together with Monte Carlo dropout as approximation to Bayesian inference. The proposed approach can easily be derived from frequentist temperature scaling and yields well-calibrated model uncertainty as well as softmax likelihood.

* Accepted at 4th workshop on Bayesian Deep Learning (NeurIPS 2019)  
View paper onarxiv icon

Share this with someone who'll enjoy it:

Twitter IconFacebook IconLinkedin IconWhatsapp IconMessenger Icon