Diverse Ensembles Improve Calibration

Asa Cooper Stickland, Iain Murray

Research output: Contribution to conferencePaperpeer-review

Abstract / Description of output

Modern deep neural networks can produce badly calibrated predictions, especially when train and test distributions are mismatched. Training an ensemble of models and averaging their predictions can help alleviate these issues. We propose a simple technique to improve calibration, using a different data augmentation for each ensemble member. We additionally use the idea of ‘mixing’unaugmented and augmented inputs to improve calibration when test and training distributions are the same. These simple techniques improve calibration and accuracy over strong baselines on the CIFAR10 and CIFAR100 benchmarks, and out-of-domain data from their corrupted versions.
Original languageEnglish
Number of pages6
Publication statusPublished - 17 Jul 2020
EventICML 2020 Workshop on Uncertainty & Robustness in Deep Learning - Virtual workshop
Duration: 17 Jul 202017 Jul 2020


WorkshopICML 2020 Workshop on Uncertainty & Robustness in Deep Learning
Abbreviated titleICML UDL 2020
CityVirtual workshop
Internet address


Dive into the research topics of 'Diverse Ensembles Improve Calibration'. Together they form a unique fingerprint.

Cite this