AI Paper Atlas

Research map / Label-efficient and robust learning / Semi-supervised learning

In Defense of Pseudo-Labeling: An Uncertainty-Aware Pseudo-label Selection Framework for Semi-Supervised Learning

Mamshad Nayeem Rizve, Kevin Duarte, Yogesh S Rawat, Mubarak Shah

ICLR 2021 · Label-efficient and robust learning › Semi-supervised learning · approach: Confidence-based pseudo-label selection · 72 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
select pseudo-labels using confidence and prediction uncertainty (ablation)pseudo-labeling on CIFAR-10 · CIFAR-10, 1000 labelserror rate (%): 16.50 -> 8.14Improved
select pseudo-labels using confidence and prediction uncertainty (ablation)pseudo-labeling on CIFAR-10 · CIFAR-10, 4000 labelserror rate (%): 10.02 -> 6.36Improved
add negative learning for selected negative pseudo-labels (ablation)UPS on CIFAR-10 · CIFAR-10, 1000 labelserror rate (%): 9.46 -> 8.14Improved
add negative learning for selected negative pseudo-labels (ablation)UPS on CIFAR-10 · CIFAR-10, 4000 labelserror rate (%): 6.64 -> 6.36Improved
add confidence-based pseudo-label selection (ablation)pseudo-labeling on CIFAR-10 · CIFAR-10, 1000 labelserror rate (%): 22.60 -> 16.50Improved
add confidence-based pseudo-label selection (ablation)pseudo-labeling on CIFAR-10 · CIFAR-10, 4000 labelserror rate (%): 12.94 -> 10.02Improved
add MC-Dropout calibration while retaining confidence selection (ablation)UPS on CIFAR-10 · CIFAR-10, 1000 labelserror rate (%): 16.50 -> 13.68Improved
add MC-Dropout calibration while retaining confidence selection (ablation)UPS on CIFAR-10 · CIFAR-10, 4000 labelserror rate (%): 10.02 -> 8.09Improved
add Mixup augmentation without negative learning (ablation)UPS on CIFAR-10 · CIFAR-10, 1000 labelserror rate (%): 9.46 -> 7.30Improved
replace MC-Dropout uncertainty estimation with MC-SpatialDropout (ablation)UPS on CIFAR-10 · CIFAR-10, 1000 labelserror rate (%): 8.14 -> 8.28No clear effect
replace MC-Dropout uncertainty estimation with MC-DropBlock (ablation)UPS on CIFAR-10 · CIFAR-10, 1000 labelserror rate (%): 8.14 -> 9.76Hurt
add class balancing during pseudo-label selection (ablation)UPS · CIFAR-10, 1000 labelsaccuracy: 88.77 -> 91.86Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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