AI Paper Atlas

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

Semi-Supervised Active Learning With Temporal Output Discrepancy

Siyu Huang, Tianyang Wang, Haoyi Xiong, Jun Huan, Dejing Dou

ICCV 2021 · Label-efficient and robust learning › Semi-supervised learning · approach: Pseudo-labeling for specialized tasks · 74 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 TOD semi-supervised task learning (ablation)Base · Cifar10 (40% labels)accuracy: 91.8 -> 92.2 (+0.4)Improved
add TOD active data selection (ablation)Base · Cifar10 (40% labels)accuracy: 91.8 -> 94.2 (+2.4)Improved
add both semi-supervised learning and active selection (ablation)Base · Cifar10 (40% labels)accuracy: 91.8 -> 94.5 (+2.7)Improved
add TOD semi-supervised task learning (ablation)Base · Cifar100 (40% labels)accuracy: 62.3 -> 66.1 (+3.8)Improved
add TOD active data selection (ablation)Base · Cifar100 (40% labels)accuracy: 62.3 -> 67.3 (+5.0)Improved
add both semi-supervised learning and active selection (ablation)Base · Cifar100 (40% labels)accuracy: 62.3 -> 68.5 (+6.2)Improved
add TOD-based unsupervised loss (ablation)task model · Cityscapes (40% labels, 19 classes)mIoU: 54.7 -> 58.9Improved
use CyclicOD (last-cycle model baseline) vs EMAOD for sampling (ablation)TOD sampling · Cifar-100accuracyMixed
use TOD-based sampling instead of random or LL4AL sampling (ablation)active learning · Cifar-10 and Cifar-100accuracyImproved
use EMA baseline model (EMAOD) for the unsupervised loss instead of CyclicOD (ablation)TOD unsupervised loss · Cifar-10 and Cifar-100accuracyImproved
replace learned loss predictor (LL4AL) with COD for samplingactive sampling · Cifar-10sampling time (s): 7.7 -> 7.2Improved

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