Research map / Label-efficient and robust learning / Recognition and long-tail
AllGCD: Leveraging All Unlabeled Data for Generalized Category Discovery
Xinzi Cao, Ke Chen, Feidiao Yang, Xiawu Zheng, Yonghong Tian, Yutong Lu
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 Intra-CL to SimGCD (ablation) | SimGCD · CUB | All ACC: 60.3 -> 65.2 | Improved |
| add Inter-CU to SimGCD (ablation) | SimGCD · CUB | All ACC: 60.3 -> 67.1 | Improved |
| combine Intra-CL and Inter-CU (ablation) | AllGCD · CUB | All ACC: 67.1 -> 68.4 | Improved |
| add Intra-CL to SimGCD (ablation) | SimGCD · CIFAR100 | New ACC: 77.8 -> 79.6 | Improved |
| add Inter-CU to SimGCD (ablation) | SimGCD · CIFAR100 | New ACC: 77.8 -> 81.7 | Improved |
| raise Intra-CL confidence from 0.75 to 0.90 (ablation) | AllGCD · CUB | All ACC: 68.4 -> 66.2 | Hurt |
| raise Intra-CL confidence from 0.70 to 0.85 (ablation) | AllGCD · CIFAR100 | All ACC: 80.8 -> 82.3 | Improved |
| raise Inter-CU threshold from 0.65 to 0.80 (ablation) | AllGCD · Stanford Cars | All ACC: 60.5 -> 57.6 | Hurt |
| raise Inter-CU threshold from 0.60 to 0.70 (ablation) | AllGCD · ImageNet-100 | All ACC: 85.6 -> 86.5 | Improved |
| use AllGCD instead of SimGCD with ground-truth K | AllGCD · Stanford Cars | New ACC: 45.0 -> 52.5 | Improved |
| start Inter-CU at later epoch (e=80 vs earlier) for hard fine-grained sets (ablation) | AllGCD · Stanford Cars/FGVC-Aircraft | New ACC | Improved |
| use AllGCD vs CMS on Old ACC | AllGCD · CIFAR100 | Old ACC: 85.7 -> 82.3 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SPTNet — SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning (ICLR 2024) · better on 5 of 6 shared results
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 0.89)
- Active Generalized Category Discovery (CVPR 2024 · similarity 0.89)
- Generalized Category Discovery via Reciprocal Learning and Class-Wise Distribution Regularization (ICML 2025 · similarity 0.89)
- Solving the Catastrophic Forgetting Problem in Generalized Category Discovery (CVPR 2024 · similarity 0.89)
- Generalized Category Discovery under Domain Shift: A Frequency Domain Perspective (NeurIPS 2025 · similarity 0.89)
- Towards Understanding Parametric Generalized Category Discovery on Graphs (ICML 2025 · similarity 0.89)
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.
