AI Paper Atlas

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

Amortized Conditional Normalized Maximum Likelihood: Reliable Out of Distribution Uncertainty Estimation

Aurick Zhou, Sergey Levine

ICML 2021 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Single-model predictive uncertainty · 6 citations (OpenAlex)

Read the paper

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 ACNML rather than Bayesian model averaging with the same posteriorBayes-by-Backprop on rotated MNIST · Rotated MNIST, severe rotationsECEImproved
Use ACNML rather than Bayesian model averaging with the same posteriorBayes-by-Backprop on rotated MNIST · MNIST in-distribution and lowest rotation levelsECEHurt
Use ACNML rather than MAP and Bayesian baselinesVGG16 on CIFAR10-C · CIFAR10-C, more severe corruptionsECEImproved
Use ACNML rather than MAP and Bayesian baselinesWideResNet28x10 on CIFAR10-C · CIFAR10-C, heavier corruptionsECEImproved
Use ACNML rather than the corresponding Bayesian methodsVGG16 and WideResNet28x10 on CIFAR10-C · CIFAR10-C, more severe corruptionsBrier scoreImproved
Use ACNML rather than the corresponding Bayesian methodsVGG16 and WideResNet28x10 on CIFAR10-C · CIFAR10-C, all corruption severitiesaccuracyNo clear effect
Use ACNML rather than naive CNMLMNIST MLP · MNIST inferenceinference time per input: 13.83s -> 0.08sImproved
Use ACNML rather than naive CNMLVGG16 · CIFAR10 inferenceinference time per input: 102.0s -> 0.37sImproved
Use ACNML rather than naive CNMLWRN28x10 · CIFAR10 inferenceinference time per input: 359.1s -> 1.1sImproved
Use ACNML with KFAC-Laplace instead of diagonal SWAG posterior (ablation)VGG16 on CIFAR10 · CIFAR10 testECE: 0.0115±0.0010 -> 0.0361± 0.0016Hurt
Use ACNML-SWAGD instead of SWAGD Bayesian model averaging (ablation)VGG16 on CIFAR10 · CIFAR10 testECE: 0.0284±0.0002 -> 0.0115±0.0010Improved
Use ACNML-SWAGD instead of SWAGD Bayesian model averaging (ablation)WideResNet28x10 on CIFAR10 · CIFAR10 testECE: 0.0042 ± 0.0003 -> 0.0122 ± 0.0006Hurt

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