AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Algorithm Selection for Deep Active Learning with Imbalanced Datasets

Jifan Zhang, Shuai Shao, saurabh verma, Robert D Nowak

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Deep active learning · 5 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
adaptive Thompson-sampling selection over candidate algorithms instead of a single best algorithmTAILOR · CelebA multi-label classificationmean average precisionImproved
use TAILOR instead of the best single baseline algorithmdeep active learning · ten imbalanced datasetsaccuracyNo clear effect
use TAILOR instead of the best baselinedeep active learning · ten imbalanced datasetsclass diversity (labels in rarest class)Improved
use TAILOR instead of confidence samplingdeep active learning · Caltech256balanced accuracyHurt
ALBL sampling (training-accuracy rewards) instead of Random Metameta algorithm selection · all datasetsaccuracyNo clear effect
TAILOR instead of existing meta algorithms (ALBL, Random Meta)meta algorithm selection · all ten datasetsaccuracy and class diversityImproved
TAILOR instead of best baseline for active searchmulti-label search · CelebA, COCO, VOC, Cartotal positive labelsImproved
run TAILOR rather than the slowest candidate algorithmTAILOR · all experimentsrunning time: $20\%$Improved
Modified Submodular without balanced holdout set (ablation)SIMILAR · multi-class classificationaccuracyHurt

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