AI Paper Atlas

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

Optimization-based Uncertainty Attribution Via Learning Informative Perturbations

Hanjing Wang, Bashirul Azam Biswas, Qiang Ji

ECCV 2024 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Single-model predictive uncertainty · 3 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 SAM-guided mask parameterization (ablation)Eq. (3) uncertainty attribution · C10 problematic region detectionIoU: 0.333 -> 0.648Improved
add learnable (blur) perturbation (ablation)Eq. (3) + SAM · SVHN problematic region detectionIoU: 0.709 -> 0.765Improved
add Gumbel-sigmoid reparameterization of binary mask (ablation)Eq. (3) + SAM + LP · SVHN problematic region detectionIoU: 0.765 -> 0.829Improved
add SAM segments to KernelSHAPKernelSHAP · C10 problematic region detectionIoU: 0.324 -> 0.676Improved
optimization-based mask vs best baseline for faithfulness with gradient descent refinementOurs vs KernelSHAP · C10/C100/SVHN avg URRURR: 0.770 -> 0.884Improved
optimization-based mask vs Occlusion for blurring refinementOurs vs Occlusion · C10/C100/SVHN avg URR, blurringURR: 0.528 -> 0.743Improved
vary lambda (3 -> 2.5 or 2) and tau (0.1 -> 0.07 or 0.15) (ablation)Ours · C10 problematic region detectionIoU: 0.713 -> 0.677 to 0.688No clear effect
refinement by gradient descent vs blurring for perturbation-based LIME/OcclusionLIME and Occlusion · URR faithfulnessURR: Occlusion avg 0.587 (gradient) vs 0.528 (blurring)Mixed
Ours vs UA-Backprop on PartImageNet ADAOurs · PartImageNet detectionADA: 0.695 -> 0.635No clear effect

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