AI Paper Atlas

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

Detecting Misclassification Errors in Neural Networks with a Gaussian Process Model

Xin Qiu, Risto Miikkulainen

AAAI 2022 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Single-model predictive uncertainty · 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 RED instead of MCP confidenceStandard NN classifier · 125 UCI datasetsAP-Error mean rank: 2.93± 0.89 -> 1.39±0.61Improved
Use RED instead of MCP confidenceStandard NN classifier · 125 UCI datasetsAUPR-Error mean rank: 3.06±0.92 -> 1.49±0.78Improved
Use RED instead of Trust ScoreVGG16 · CIFAR-10AP-Error (%): 48.76±2.28 -> 49.88±1.99Improved
Use RED instead of Trust ScoreWRN-10-4 · CIFAR-10AP-Error (%): 51.15±3.29 -> 52.51±2.81Improved
Use RED instead of Trust ScoreVGG19 · CIFAR-100AP-Error (%): 72.63±1.20 -> 73.40±1.05Improved
Replace the GP with Bayesian linear regression (ablation)RED · 125 UCI datasetsAP-Error pairwise comparisonHurt
Use RED-variance rather than MCPOOD detection · SVHN OOD vs CIFAR-10, VGG16AP-OOD (%): 82.964±1.850 -> 86.282±2.212Improved
Use RED-variance rather than RED-meanRED · 125 UCI datasets with OOD and adversarial samplesAP-OOD and AP-AdversarialImproved
Apply RED to the BNN or MC-Dropout classifier rather than using its internal confidence scoreBNN and MC-Dropout classifiers · 125 UCI datasetsAP-Error pairwise comparisonsImproved
Use RED rather than MCP when errors are few or classifier accuracy is highStandard NN classifier · 125 UCI datasets, few errors or high accuracyAP-ErrorImproved

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