AI Paper Atlas

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

Understanding Contrastive Learning via Distributionally Robust Optimization

Junkang Wu, Jiawei Chen, Jiancan Wu, Wentao Shi, Xiang Wang, Xiangnan He

NeurIPS 2023 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 12 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
Replace InfoNCE with ADNCESimCLR, ResNet50 · CIFAR10, 400 epochsTop-1 accuracy: 91.64 -> 91.88Improved
Replace InfoNCE with ADNCESimCLR, ResNet50 · STL10, 400 epochsTop-1 accuracy: 87.69 -> 88.00Improved
Replace InfoNCE with ADNCESimCLR, ResNet50 · CIFAR100, 400 epochsTop-1 accuracy: 69.03 -> 69.35Improved
Replace InfoNCE with ADNCESimCSE-BERTbase · STS averageSpearman correlation: 76.81 -> 77.03Improved
Replace InfoNCE with ADNCESimCSE-RoBERTabase · STS averageSpearman correlation: 76.81 -> 77.10Improved
Tune temperature τ by grid search instead of using τ₀SimCLR, ResNet50 · STL10, 400 epochsTop-1 accuracy: 81.67 -> 87.69Improved
Tune temperature τ by grid search instead of using τ₀SimCLR, ResNet50 · CIFAR100, 400 epochsTop-1 accuracy: 66.53 -> 69.03Improved
Tune temperature τ by grid search instead of using τ₀SimCSE-BERTbase · STS averageSpearman correlation: 76.25 -> 76.81Improved
Tune temperature τ by grid search instead of using τ₀SimCSE-RoBERTabase · STS averageSpearman correlation: 76.57 -> 76.81Improved
Replace InfoNCE with the mean-variance objective (ablation)Contrastive learning model · Reported benchmark experimentsPerformanceHurt

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