Research map / Label-efficient and robust learning / Self-supervised and contrastive
Prototype-based Dataset Comparison
Nanne van Noord
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 |
|---|---|---|---|
| increase number of prototypes K from 1024 to 8192 (ablation) | ProtoSim · ImageNet val linear probe | accuracy: 54.7 -> 63.7 -> 64.8 (K=1024, 4096, 8192) | Improved |
| train ProtoSim on ImageNet+PASS union instead of ImageNet only (ablation) | ProtoSim · ImageNet val linear probe | accuracy: 64.8 -> 53.2 | Hurt |
| use PASS-pretrained backbone instead of ImageNet-pretrained DINO (ablation) | ProtoSim · ImageNet val linear probe | accuracy: 64.8 -> 20.4 | Hurt |
| zero out the most frequent prototype of each class (ablation) | ProtoSim on ImageNet · ImageNet per-class accuracy (top 100 classes) | accuracy drop: 22.6 percentage points average difference | Hurt |
| jointly learn prototypes across datasets to compare PASS and ImageNet | ProtoSim · PASS vs ImageNet case study | qualitative | Improved |
Closest papers by meaning
- Hunting Attributes: Context Prototype-Aware Learning for Weakly Supervised Semantic Segmentation (CVPR 2024 · similarity 0.83)
- ConceptScope: Characterizing Dataset Bias via Disentangled Visual Concepts (NeurIPS 2025 · similarity 0.83)
- Mind the Gap Between Prototypes and Images in Cross-domain Finetuning (NeurIPS 2024 · similarity 0.83)
- Multi-Prototype Space Learning for Commonsense-Based Scene Graph Generation (AAAI 2024 · similarity 0.83)
- Nonparametric Identification of Latent Concepts (ICML 2025 · similarity 0.83)
- ConceptBed: Evaluating Concept Learning Abilities of Text-to-Image Diffusion Models (AAAI 2024 · similarity 0.83)
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.
