Research map / Label-efficient and robust learning / Recognition and long-tail
ProtoGCD: Unified and Unbiased Prototype Learning for Generalized Category Discovery
Shijie Ma, Fei Zhu, Xu-Yao Zhang, Cheng‐Lin Liu
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 DAPL pseudo-labeling to contrastive learning (ablation) | ProtoGCD · CIFAR100 | All accuracy: 64.0 -> 65.8 | Improved |
| add entropy regularization without DAPL (ablation) | ProtoGCD · CIFAR100 | All accuracy: 64.0 -> 30.1 | Hurt |
| add entropy regularization on top of DAPL (ablation) | ProtoGCD · CIFAR100 | All accuracy: 65.8 -> 79.7 | Improved |
| add entropy regularization (with separation) (ablation) | ProtoGCD · FGVC-Aircraft | All accuracy: 38.0 -> 56.8 | Improved |
| add inter-class separation regularization (ablation) | ProtoGCD · CIFAR100 | All accuracy: 79.7 -> 81.9 | Improved |
| add inter-class separation regularization (ablation) | ProtoGCD · FGVC-Aircraft | All accuracy: 54.4 -> 56.8 | Improved |
| add marginal entropy maximization on long-tailed Herbarium19 (ablation) | ProtoGCD · Herbarium19 | All accuracy: 29.4 -> 44.5 | Improved |
| add small entropy weight 0.1 on Herbarium19 (ablation) | ProtoGCD · Herbarium19 | All accuracy: 29.4 -> 36.2 | Improved |
| add sharpening of soft pseudo-labels (ablation) | ProtoGCD DAPL · CIFAR100 and Aircraft | All accuracy: ~20% | Improved |
| use adaptive hard/soft ratio instead of fixed 1:1 (ablation) | ProtoGCD DAPL · CIFAR100 and Aircraft | All accuracy: ~2% lower with fixed 1:1 | Improved |
| vary ramp-up epochs e_ramp, optimum near 125 (ablation) | ProtoGCD · CIFAR100 and Aircraft | All accuracy: stable within [75,150] | No clear effect |
| use DINOv2 initialization instead of DINO | ProtoGCD · CUB | All accuracy: 63.2 -> 74.9 | Improved |
Later papers that compared with this method
3 later accepted papers report ProtoGCD as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- Open-World Deepfake Attribution via Confidence-Aware Asymmetric Learning (AAAI 2026 · better on 3 of 3 shared results)
- Learning Intrinsic Hierarchy for Generalized Category Discovery (AAAI 2026 · better on 2 of 2 shared results)
- Foundation-Adaptive Integrated Refinement for Generalized Category Discovery (AAAI 2026 · 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
- DebGCD: Debiased Learning with Distribution Guidance for Generalized Category Discovery (ICLR 2025 · similarity 0.91)
- Generalized Category Discovery via Reciprocal Learning and Class-Wise Distribution Regularization (ICML 2025 · similarity 0.90)
- Active Generalized Category Discovery (CVPR 2024 · similarity 0.90)
- Prior-Constrained Association Learning for Fine-Grained Generalized Category Discovery (AAAI 2025 · similarity 0.90)
- Happy: A Debiased Learning Framework for Continual Generalized Category Discovery (NeurIPS 2024 · similarity 0.90)
- Transfer and Alignment Network for Generalized Category Discovery (AAAI 2024 · similarity 0.90)
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.
