AI Paper Atlas

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

Learning Efficient Coding of Natural Images with Maximum Manifold Capacity Representations

Thomas Edward Yerxa, Yilun Kuang, Eero P Simoncelli, SueYeon Chung

NeurIPS 2023 · Label-efficient and robust learning › Self-supervised and contrastive · approach: Invariance-based self-supervised 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
reduce batch size from 2048 to 256 (ablation)MMCR · ImageNet-1k linear evalaccuracy: 1.2% reductionHurt
add a momentum encoder for ImageNet-1k pre-training (ablation)MMCR · ImageNet-1k linear evalaccuracy: +1%Improved
increase number of views (third MMCR row vs first) (ablation)MMCR · ImageNet-1k linear evalaccuracy: 69.5 -> 72.1Improved
increase number of views (third MMCR row vs first) (ablation)MMCR · ImageNet-1k 1% labelsaccuracy: 46.6 -> 51.0Improved
increase number of views (third MMCR row vs first) (ablation)MMCR · Flowers-102accuracy: 90.0 -> 91.4Improved
MMCR vs SwAV on transfer to Food-101 and DTDSwAV · Food-101accuracy: 72.1 -> 73.6Improved
MMCR vs SwAV on ImageNet linear evalSwAV · ImageNet-1k linear evalaccuracy: 72.1 -> 72.1No clear effect
MMCR vs baselines on VOC07 detectionMMCR · VOC07 detectionmAP: 54.6No clear effect
MMCR instead of SimCLR for neural predictivity of V2SimCLR · BrainScore V2correlation: .288 -> .311Improved
MMCR instead of SwAV for neural predictivity of ITSwAV · BrainScore ITcorrelation: .398 -> .416Improved
MMCR instead of SimCLR for V1 neural predictivitySimCLR · BrainScore V1correlation: .500 -> .494No clear effect
MMCR instead of SimCLR for representation dimensionalitySimCLR · ImageNet valparticipation ratio: 124.6 -> 279.2Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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