AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

Contextual Similarity Distillation: Ensemble Uncertainties with a Single Model

Moritz Akiya Zanger, Pascal R. Van der Vaart, Wendelin Boehmer, Matthijs T. J. Spaan

ICLR 2026 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Epistemic versus aleatoric decomposition · 6 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add data augmentations as context data (ablation)CSD · MNIST-family distribution shiftAUROC: 96.63 -> 98.22Improved
add unlabeled target-distribution context data (ablation)CSD · MNIST-family distribution shiftAUROC: 96.63 -> 98.57Improved
add target-distribution context data on top of augmentations (ablation)CSD-Aug. · MNIST-family distribution shiftAUROC: 98.22 -> 98.57Improved
add augmentations to context data (ablation)CSD · MNIST-family distribution shiftAUPR-OUT: 96.19 -> 97.80Improved
add target-distribution context data (ablation)CSD · MNIST-family distribution shiftAUPR-IN: 96.94 -> 98.86Improved
add augmentations to context data (ablation)CSD · MNIST-family distribution shifttest accuracy: 96.29 -> 96.28No clear effect
single-model CSD instead of 15-member deep ensembleENS(15) · MNIST-family distribution shiftAUROC: 94.00 -> 96.63Improved
single-model CSD instead of RNDRND · MNIST-family distribution shiftAUROC: 94.40 -> 96.63Improved
single-model CSD instead of ensemble of 15 on accuracyENS(15) · MNIST-family distribution shifttest accuracy: 97.18 -> 96.29Hurt
increase ensemble size from 3 to 15 (ablation)deep ensemble · MNIST-family distribution shiftAUROC: 92.30 -> 94.00Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy