AI Paper Atlas

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

Monotonic Variational Gaussian Process for Efficient Data Collection

Donghyun Lee, Young Myoung Ko

ICML 2026 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: PAC-Bayes and decoupled GP learning

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
replace power-law LOC with monotonic variational GP MOVEdata collection planner · CIFAR-10 (T=3)cost ratio: 0.0568 -> 0.0433Improved
replace power-law LOC with MOVEdata collection planner · CIFAR-100 (T=3)cost ratio: 0.0633 -> 0.0404Improved
replace power-law LOC with MOVEdata collection planner · VOC-Seg (T=3)cost ratio: 0.1516 -> 0.1045Improved
replace power-law LOC with MOVEdata collection planner · ImageNetaccuracy ratio: -0.03 -> 0.02Improved
replace LOC with MOVEdata collection planner · VOC-Det (T=5)cost ratio: 0.1159 -> 0.0614Improved
replace LOC with MOVEdata collection planner · VOC-Det (T=3)cost ratio: 0.1538 -> 0.0633Hurt
add monotonicity constraint (MOVE instead of ExactGP) (ablation)expected shortfall planner · CIFAR-10 (T=3)cost ratio: 0.0640 -> 0.0433Improved
add monotonicity constraint (MOVE instead of ExactGP) (ablation)expected shortfall planner · CIFAR-100 (T=5)cost ratio: 0.0592 -> 0.0334Improved
add monotonicity constraint (MOVE instead of ExactGP) (ablation)expected shortfall planner · VOC-Seg (T=5)cost ratio: 0.1465 -> 0.1041Improved
add monotonicity constraint (MOVE instead of ExactGP) (ablation)expected shortfall planner · VOC-Det (T=3)cost ratio: 0.0587 -> 0.0633Hurt
use expected shortfall instead of failure probability as terminal penalty (ablation)MOVE · CIFAR-10 (T=1)penalty gradient and success probabilityImproved
use monotonic variational GP instead of SVGP (ablation)learning curve model · Synthetic power-law extrapolationMSE: 1127.60 -> 700.99Improved

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