Research map / Label-efficient and robust learning / Recognition and long-tail
The Devil Is in Gradient Entanglement: Energy-Aware Gradient Coordinator for Robust Generalized Category Discovery
Haiyang Zheng, Nan Pu, Yaqi Cai, Teng Long, Wenjing Li, Nicu Sebe, Zhun Zhong
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 Anchor-based Gradient Alignment (AGA) (ablation) | SimGCD · CUB | All ACC: 60.3 -> 65.6 | Improved |
| add Anchor-based Gradient Alignment (AGA) (ablation) | SelEx · CUB | All ACC: 73.6 -> 81.7 | Improved |
| add Energy-aware Elastic Projection (EEP) (ablation) | SimGCD · CUB | All ACC: 60.3 -> 65.2 | Improved |
| add Energy-aware Elastic Projection (EEP) (ablation) | SelEx · CUB | All ACC: 73.6 -> 82.3 | Improved |
| add EEP on top of AGA (full EAGC) (ablation) | SimGCD + AGA · Stanford Cars | All ACC: 62.2 -> 62.9 | Improved |
| add AGA on top of EEP (full EAGC) (ablation) | SelEx + EEP · Stanford Cars | All ACC: 58.8 -> 65.7 | Improved |
| use energy-aware weighting instead of uniform projection (ablation) | SelEx + EEP · CUB | All ACC: 80.9 -> 83.2 | Improved |
| use Conceptor soft subspace instead of PCA-16 (ablation) | SelEx + EEP · CUB | All ACC: 82.7 -> 83.2 | Improved |
| plug EAGC into SimGCD | SimGCD · CUB | All ACC: 60.3 -> 66.5 | Improved |
| plug EAGC into SelEx | SelEx · CUB | All ACC: 73.6 -> 83.2 | Improved |
| plug EAGC into SelEx | SelEx · CIFAR-100 | All ACC: 82.3 -> 79.3 | Hurt |
| plug EAGC into SimGCD | SimGCD · CUB | GDC (gradient deviation): 0.2669 -> 0.1525 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- SelEx — SelEx: Self-Expertise in Fine-Grained Generalized Category Discovery (ECCV 2024) · better on 9 of 9 shared results
- SPTNet — SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning (ICLR 2024) · 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
- Consistent Supervised-Unsupervised Alignment for Generalized Category Discovery (NeurIPS 2025 · similarity 0.87)
- Happy: A Debiased Learning Framework for Continual Generalized Category Discovery (NeurIPS 2024 · similarity 0.86)
- Continual Generalized Category Discovery: Learning and Forgetting from a Bayesian Perspective (ICML 2025 · similarity 0.86)
- HIDISC: A Hyperbolic Framework for Domain Generalization with Generalized Category Discovery (NeurIPS 2025 · similarity 0.86)
- Active Generalized Category Discovery (CVPR 2024 · similarity 0.86)
- DebGCD: Debiased Learning with Distribution Guidance for Generalized Category Discovery (ICLR 2025 · similarity 0.85)
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.
