AI Paper Atlas

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

On Uncertainty, Tempering, and Data Augmentation in Bayesian Classification

Sanyam Kapoor, Wesley Maddox, Pavel Izmailov, Andrew Gordon Wilson

NeurIPS 2022 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Epistemic versus aleatoric decomposition · 9 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 a cold tempered softmax likelihood instead of the standard softmax likelihoodResNet-18 Bayesian neural network · CIFAR-10test accuracyImproved
Use the Noisy Dirichlet model instead of the standard softmax likelihoodResNet-18 Bayesian neural network · CIFAR-10test accuracyImproved
Model label noise with likelihood tempering or the Noisy Dirichlet modelResNet-18 Bayesian neural network · CIFAR-10 and Tiny ImageNet with varying label noisetest accuracyImproved
Use the Noisy Dirichlet model rather than tempered softmaxResNet-18 Bayesian neural network · Tiny ImageNet with high label noisetest accuracyImproved
Add data augmentation without compensating for its softened likelihood (ablation)Bayesian neural network · CIFAR-10 train datanegative log-likelihoodHurt
Add likelihood tempering to counteract data augmentation (ablation)Bayesian neural network · Synthetic two-spirals classificationtraining fitImproved
Use the Noisy Dirichlet model instead of tempering (ablation)Bayesian neural network · Synthetic two-spirals classificationtraining fitImproved
Use the Noisy Dirichlet model with posterior temperature 1 rather than adding posterior tempering (ablation)ResNet-18 Bayesian neural network · CIFAR-10test accuracyNo clear effect
Use more complex data augmentation and lower the posterior temperature (ablation)ResNet-18 Bayesian neural network · CIFAR-10test negative log-likelihoodMixed
Use a proper augmentation likelihood rather than the standard likelihood (ablation)Bayesian neural network · Image classificationperformanceNo clear effect
Use a smoothed softmax likelihood rather than a tempered likelihood (ablation)ResNet-18 Bayesian neural network · CIFAR-10test performanceHurt
Vary the prior at temperature 1 rather than tempering (ablation)Bayesian neural network · CIFAR-10test accuracyHurt

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