Well-calibrated Model Uncertainty with Temperature Scaling for Dropout Variational Inference
This work addresses uncertainty calibration for practitioners using dropout-based Bayesian neural networks, but it is incremental as it adapts an existing frequentist method to a specific inference setting.
The paper tackles the problem of miscalibrated model uncertainty in variational Bayesian inference with Monte Carlo dropout by extending temperature scaling to calibrate it, resulting in a significant reduction in uncertainty calibration error (UCE) on CIFAR-10/100 datasets without affecting model accuracy.
Model uncertainty obtained by variational Bayesian inference with Monte Carlo dropout is prone to miscalibration. The uncertainty does not represent the model error well. In this paper, temperature scaling is extended to dropout variational inference to calibrate model uncertainty. Expected uncertainty calibration error (UCE) is presented as a metric to measure miscalibration of uncertainty. The effectiveness of this approach is evaluated on CIFAR-10/100 for recent CNN architectures. Experimental results show, that temperature scaling considerably reduces miscalibration by means of UCE and enables robust rejection of uncertain predictions. The proposed approach can easily be derived from frequentist temperature scaling and yields well-calibrated model uncertainty. It is simple to implement and does not affect the model accuracy.