AI Paper Atlas

Research map / Label-efficient and robust learning / Self-supervised and contrastive

An Augmentation Overlap Theory of Contrastive Learning

Qi Zhang, Yifei Wang, Yisen Wang

JMLR 2025 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Inductive bias and minimal representations · 1 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
increase augmentation strength r from none to moderate on synthetic data (ablation)InfoNCE contrastive learning · Synthetic two-class sphere datalinear accuracy: accuracy increases first and then falls to 50%Mixed
use appropriate RandomResizedCrop strength instead of too weak (0.99,1) (ablation)SimCLR · CIFAR-10linear accuracy: 25% -> 75%Improved
make RandomResizedCrop too strong instead of appropriate (ablation)SimCLR · CIFAR-10linear accuracy: 75% -> 29%Hurt
use GARC instead of rotation accuracy as unsupervised representation metricrepresentation evaluation · CIFAR-10Pearson correlation with linear accuracy: 0.905 -> 0.954Improved
use InfoNCE loss as indicator of downstream performancerepresentation evaluation · CIFAR-10InfoNCE loss vs linear accuracy: InfoNCE 4.8 to 5.0 while accuracy 40% to 80%No clear effect
replace conditional independence assumption with augmentation overlap assumption in the boundcontrastive learning theory · Upper bound of supervised lossbound error as M grows: L_unsup(f) + e/sqrt(M)Improved

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