Research map / Label-efficient and robust learning / Self-supervised and contrastive
GATCluster: Self-Supervised Gaussian-Attention Network for Image Clustering
Chuang Niu, Jun Zhang, Ge Wang, Jimin Liang
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 |
|---|---|---|---|
| add attention mapping module (ablation) | GATCluster · STL10 | best ACC: 0.539 -> 0.583 | Improved |
| add attention mapping module (ablation) | GATCluster · STL10 | mean ACC: 0.494 -> 0.537 | Improved |
| use grayscale instead of color images (ablation) | GATCluster · STL10 | best ACC: 0.556 -> 0.583 | Improved |
| add transformation invariance task (ablation) | GATCluster · STL10 | best ACC: 0.576 -> 0.583 | Improved |
| add separability maximization task (ablation) | GATCluster · STL10 | best ACC: 0.579 -> 0.583 | Improved |
| add separability maximization task (ablation) | GATCluster · STL10 | mean ACC: 0.529 -> 0.537 | Improved |
| add transformation invariance task (ablation) | GATCluster · STL10 | mean ACC: 0.546 -> 0.537 | No clear effect |
| add entropy analysis task (ablation) | GATCluster · STL10 | clustering | Improved |
| increase input size from 96x96 to 128x128 (ablation) | GATCluster · ImageNet-10 | best ACC: 0.739 -> 0.762 | Improved |
| increase input size from 128x128 to 160x160 (ablation) | GATCluster · ImageNet-10 | best ACC: 0.762 -> 0.712 | Hurt |
| increase input size from 128x128 to 192x192 (ablation) | GATCluster · ImageNet-10 | mean ACC: 0.735 -> 0.608 | Hurt |
| use 6x6 attention map instead of 2x2 (ablation) | GATCluster · ImageNet-10 | best ACC: 0.746 -> 0.762 | Improved |
Closest papers by meaning
- Image Clustering via the Principle of Rate Reduction in the Age of Pretrained Models (ICLR 2024 · similarity 0.88)
- Semi-Supervised Learning Made Simple With Self-Supervised Clustering (CVPR 2023 · similarity 0.87)
- Attention-based clustering (NeurIPS 2025 · similarity 0.86)
- Clustering via Self-Supervised Diffusion (ICML 2025 · similarity 0.86)
- Image Clustering with External Guidance (ICML 2024 · similarity 0.86)
- Clustering Properties of Self-Supervised Learning (ICML 2025 · similarity 0.85)
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.
