AI Paper Atlas

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

Quantification of Uncertainty with Adversarial Models

Kajetan Schweighofer, Lukas Aichberger, Mykyta Ielanskyi, Günter Klambauer, Sepp Hochreiter

NeurIPS 2023 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Epistemic versus aleatoric decomposition · 5 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
use QUAM instead of Deep Ensembles epistemic uncertaintyuncertainty estimation · MNIST vs FMNISTAUROC: .988 -> .994Improved
use QUAM instead of Deep Ensemblesuncertainty estimation · MNIST vs EMNISTAUROC: .924 -> .937Improved
use QUAM instead of Deep Ensemblesuncertainty estimation · MNIST vs OMNIGLOTAUROC: .983 -> .992Improved
use cSG-HMC, Laplace or MC dropout epistemic uncertainty instead of reference aleatoric uncertaintyreference LeNet · MNIST vs FMNISTAUROC: .986 -> .977Hurt
use QUAM instead of Deep Ensembles (all)EfficientNet · ImageNet-OAUROC: .709 -> .753Improved
use QUAM instead of Deep Ensembles (all)EfficientNet · ImageNet-AAUROC: .874 -> .872No clear effect
use QUAM instead of reference aleatoric uncertaintyEfficientNet · ImageNet-1K misclassificationAUROC: .867 -> .904Improved
use QUAM instead of reference aleatoric uncertaintyEfficientNet · ImageNet-1K selective predictionAUC: .958 -> .969Improved
use DE (LL) last-layer ensemble instead of reference aleatoric uncertaintyEfficientNet · ImageNet-OAUROC: .626 -> .562Hurt
use MC dropout instead of DE (all)EfficientNet · ImageNet-OAUROC: .709 -> .680Mixed
restrict adversarial model search to top N classes under matched compute (ablation)QUAM · ImageNet-O/AAUROCImproved
targeted per-class search instead of untargeted search (ablation)QUAM · implementationpredictive diversityImproved

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