atlas

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

Do More Negative Samples Necessarily Hurt In Contrastive Learning?

ICML 2022 · Label-efficient and robust learning › Recognition and long-tail · approach: Data augmentation and generalization bias · 10 citations (OpenAlex)

Read the paperCheck your idea against 108,967 papers

AI-extracted summary. Written by AI from the paper's title only (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 the number of negative samples from k=1 to k=512Contrastive learning on CIFAR-10 · CIFAR-10downstream classification accuracyHelped
Increase the number of negative samples kPopulation NCE-optimal representation with non-overlapping, uniformly distributed latent classes · Uniform latent classesdownstream supervised lossNo clear effect
Increase the number of negative samples kNumerically computed population NCE-optimal representation · Non-uniform latent classes, C in {3, 9}downstream supervised lossMixed
Train with CIFAR-100 data under the paper's positive-pair setupContrastive learning · CIFAR-100 testdownstream classification accuracy: 64.76%Helped

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