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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| reduce batch size from 2048 to 256 (ablation) | MMCR · ImageNet-1k linear eval | accuracy: 1.2% reduction | Hurt |
| add a momentum encoder for ImageNet-1k pre-training (ablation) | MMCR · ImageNet-1k linear eval | accuracy: +1% | Improved |
| increase number of views (third MMCR row vs first) (ablation) | MMCR · ImageNet-1k linear eval | accuracy: 69.5 -> 72.1 | Improved |
| increase number of views (third MMCR row vs first) (ablation) | MMCR · ImageNet-1k 1% labels | accuracy: 46.6 -> 51.0 | Improved |
| increase number of views (third MMCR row vs first) (ablation) | MMCR · Flowers-102 | accuracy: 90.0 -> 91.4 | Improved |
| MMCR vs SwAV on transfer to Food-101 and DTD | SwAV · Food-101 | accuracy: 72.1 -> 73.6 | Improved |
| MMCR vs SwAV on ImageNet linear eval | SwAV · ImageNet-1k linear eval | accuracy: 72.1 -> 72.1 | No clear effect |
| MMCR vs baselines on VOC07 detection | MMCR · VOC07 detection | mAP: 54.6 | No clear effect |
| MMCR instead of SimCLR for neural predictivity of V2 | SimCLR · BrainScore V2 | correlation: .288 -> .311 | Improved |
| MMCR instead of SwAV for neural predictivity of IT | SwAV · BrainScore IT | correlation: .398 -> .416 | Improved |
| MMCR instead of SimCLR for V1 neural predictivity | SimCLR · BrainScore V1 | correlation: .500 -> .494 | No clear effect |
| MMCR instead of SimCLR for representation dimensionality | SimCLR · ImageNet val | participation ratio: 124.6 -> 279.2 | Improved |
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.
- BarlowTwins — Barlow Twins: Self-Supervised Learning via Redundancy Reduction (ICML 2021) · better on 2 of 2 shared results
- VICReg — VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning (ICLR 2022) · better on 1 of 1 shared result
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
- Contrastive Self-Supervised Learning As Neural Manifold Packing (NeurIPS 2025 · similarity 0.86)
- MCNC: Manifold-Constrained Reparameterization for Neural Compression (ICLR 2025 · similarity 0.85)
- Data-free Neural Representation Compression with Riemannian Neural Dynamics (ICML 2024 · similarity 0.84)
- The Geometric Wall: Manifold Structure Predicts Layerwise Sparse Autoencoder Scaling Laws (NeurIPS 2026 · similarity 0.84)
- Feature Learning beyond the Lazy-Rich Dichotomy: Insights from Representational Geometry (ICML 2025 · similarity 0.84)
- CW Complex Hypothesis for Image Data (ICML 2024 · similarity 0.84)
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.
