Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Intrinsic Hierarchy for Generalized Category Discovery
Yu Duan, Junzhi He, Zhanxuan Hu, Mengda Ji, Rong Wang, Quanxue Gao
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 LEAH (pruner and constructor) to SimGCD | SimGCD · Stanford Cars (All) | clustering accuracy: 53.8 -> 65.5 | Improved |
| add LEAH to SimGCD | SimGCD · Stanford Cars (New) | clustering accuracy: 45.0 -> 58.2 | Improved |
| add LEAH to SimGCD | SimGCD · CIFAR-10 (All) | clustering accuracy: 97.1 -> 97.5 | No clear effect |
| add LEAH to SimGCD | SimGCD · CUB (Old) | clustering accuracy: 65.6 -> 76.3 | Mixed |
| add the pruner to baselines (ablation) | GCD and SimGCD · fine-grained and generic benchmarks | accuracy | Improved |
| add the constructor on top of the pruner (ablation) | LEAH · GCD benchmarks | accuracy | Improved |
| make alpha_in learnable instead of fixed at 0.9 (ablation) | SimGCD+LEAH · CUB (All) | clustering accuracy: 68.6 -> 69.4 | Improved |
| make both alpha_in and alpha_out learnable (ablation) | SimGCD+LEAH · CUB (New) | clustering accuracy: 64.5 -> 65.9 | Improved |
| fix all cone thresholds (ablation) | SimGCD+LEAH · CUB (Old) | clustering accuracy: 76.3 -> 76.5 | No clear effect |
| retain 40 instead of 20 key-object tokens (ablation) | SimGCD+LEAH · CUB (All) | clustering accuracy: 55.0 -> 69.4 | Improved |
| retain 60 instead of 40 key-object tokens (ablation) | SimGCD+LEAH · Stanford Cars (All) | clustering accuracy: 61.6 -> 65.5 | Improved |
| retain 60 instead of 40 key-object tokens (ablation) | SimGCD+LEAH · CUB (All) | clustering accuracy: 69.4 -> 67.7 | 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.
- ProtoGCD — ProtoGCD: Unified and Unbiased Prototype Learning for Generalized Category Discovery (TPAMI 2025) · better on 2 of 2 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
- SEAL: Semantic-Aware Hierarchical Learning for Generalized Category Discovery (NeurIPS 2025 · similarity 0.91)
- HiLo: A Learning Framework for Generalized Category Discovery Robust to Domain Shifts (ICLR 2025 · similarity 0.89)
- Hyperbolic Category Discovery (CVPR 2025 · similarity 0.89)
- HIDISC: A Hyperbolic Framework for Domain Generalization with Generalized Category Discovery (NeurIPS 2025 · similarity 0.88)
- Prior-Constrained Association Learning for Fine-Grained Generalized Category Discovery (AAAI 2025 · similarity 0.88)
- DebGCD: Debiased Learning with Distribution Guidance for Generalized Category Discovery (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.
