Research map / Label-efficient and robust learning / Recognition and long-tail
Generalized Category Discovery
Check 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 changed | Where | Measured | Effect |
|---|---|---|---|
| Use contrastively fine-tuned ViT features with semi-supervised k-means instead of the strongest baseline, UNO+ | GCD · CIFAR100 | ACC (All): 69.5 -> 73.0 | Helped |
| Use contrastively fine-tuned ViT features with semi-supervised k-means instead of the strongest baseline, UNO+ | GCD · CUB | ACC (All): 35.1 -> 51.3 | Helped |
| Use contrastively fine-tuned ViT features with semi-supervised k-means instead of the strongest baseline, UNO+ | GCD · Stanford Cars | ACC (All): 35.5 -> 39.0 | Helped |
| Use contrastively fine-tuned ViT features with semi-supervised k-means instead of the strongest baseline, UNO+ | GCD · Herbarium19 | ACC (All): 28.3 -> 35.4 | Helped |
| Replace ResNet-50 DINO features with ViT-B-16 DINO features (ablation) | GCD feature clustering · CIFAR100 | ACC (All): 34.0 -> 52.0 | Helped |
| Replace ResNet-50 DINO features with ViT-B-16 DINO features (ablation) | GCD feature clustering · Herbarium19 | ACC (All): 12.1 -> 12.9 | Helped |
| Add unsupervised contrastive fine-tuning to ViT DINO features (ablation) | GCD feature clustering · CIFAR100 | ACC (All): 52.0 -> 54.6 | Helped |
| Add supervised contrastive fine-tuning to ViT DINO features (ablation) | GCD feature clustering · CIFAR100 | ACC (All): 52.0 -> 60.5 | Helped |
| Combine supervised and unsupervised contrastive fine-tuning (ablation) | GCD feature clustering · CIFAR100 | ACC (All): 52.0 -> 71.1 | Helped |
| Add semi-supervised constraints to k-means after both contrastive losses (ablation) | GCD feature clustering · CIFAR100 | ACC (All): 71.1 -> 73.0 | Helped |
| Add semi-supervised constraints to k-means after both contrastive losses (ablation) | GCD feature clustering · Herbarium19 | ACC (All): 28.7 -> 35.4 | Helped |
| Add semi-supervised constraints to k-means after both contrastive losses (ablation) | GCD feature clustering · CIFAR100 | ACC (Old): 78.3 -> 76.2 | Hurt |
Closest papers by meaning
- Semantic-Guided Novel Category Discovery (AAAI 2024 · similarity 0.90)
- Solving the Catastrophic Forgetting Problem in Generalized Category Discovery (CVPR 2024 · similarity 0.89)
- SEAL: Semantic-Aware Hierarchical Learning for Generalized Category Discovery (NeurIPS 2025 · similarity 0.88)
- SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning (ICLR 2024 · similarity 0.88)
- DebGCD: Debiased Learning with Distribution Guidance for Generalized Category Discovery (ICLR 2025 · similarity 0.88)
- HiLo: A Learning Framework for Generalized Category Discovery Robust to Domain Shifts (ICLR 2025 · similarity 0.88)
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.