AI Paper Atlas

Research map / Learning theory and algorithms / Sample complexity

Can semi-supervised learning use all the data effectively? A lower bound perspective

Alexandru Tifrea, Gizem Yüce, Amartya Sanyal, Fanny Yang

NeurIPS 2023 · Learning theory and algorithms › Sample complexity · approach: Generalization and limits under data constraints · 0 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
use SSL-W (weighted SL + UL+ ensemble) instead of SL or UL+ aloneSSL-W · synthetic 2-GMM, varying SNR (nl=20, nu=2000)classification errorImproved
use SSL-W instead of switching algorithm SSL-S at moderate SNRSSL-W · synthetic 2-GMMclassification errorImproved
vary labeled set size at fixed SNR 0.5 and nu=7000SSL-W · synthetic 2-GMM, varying nu/nlclassification errorImproved
use SSL-W on real-world binary datasetsSSL-W · 10 OpenML and MNIST-pair datasetsclassification error gap vs SL and UL+Improved
use self-training with logistic regression and ridge penaltyself-training · 10 real-world binary datasetsclassification error gap vs SL and UL+Improved
self-training compared with SSL-Sself-training · 2-GMM / real dataclassification errorImproved
use labeled data in the unsupervised step as wellUL+ · 2-GMM minimax raterate improvementNo clear effect
large unlabeled set relative to SNRSSL · 2-GMMimprovement over UL+No clear effect
large labeled setSSL · 2-GMMimprovement over SLNo 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