Research map / Continual learning / Class-incremental
MetaGCD: Learning to Continually Learn in Generalized Category Discovery
Yanan Wu, Zhixiang Chi, Yang Wang, Songhe Feng
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 candidate neighbors as positives in contrastive learning (ablation) | VanillaGCD baseline · C-GCD CIFAR100 | mean All accuracy: 73.67 -> 74.92 | Improved |
| add soft positiveness weights to neighbors (ablation) | VanillaGCD + candidate neighbors · C-GCD CIFAR100 | mean New accuracy: 62.89 -> 64.55 | Improved |
| add meta-learning bi-level optimization on pseudo incremental sessions (ablation) | MetaGCD without meta-learning · C-GCD CIFAR100 | mean All accuracy: 75.66 -> 77.06 | Improved |
| use non-parametric identity mapping instead of learned projections in attention module (ablation) | soft positiveness · C-GCD CIFAR100 | accuracy | Improved |
| set neighbor similarity threshold to 0.85 rather than lower or higher (ablation) | MetaGCD · C-GCD CIFAR100 | accuracy | Mixed |
| sample more novel classes in meta-training sequences (ablation) | MetaGCD · C-GCD CIFAR100 | accuracy | Improved |
| MetaGCD versus GM on new-class discovery | GM · C-GCD Tiny-ImageNet | final New accuracy: 50.90 -> 58.46 | Improved |
| MetaGCD versus GM on old-class retention | GM · C-GCD Tiny-ImageNet | final Old accuracy: 72.08 vs 71.53 | Hurt |
Later papers that compared with this method
5 later accepted papers report MetaGCD as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.
- Virtual Category-Guided Continual Generalized Category Discovery (ECCV 2026 · better on 3 of 3 shared results)
- Decouple Your Discovery and Memory in Continual Generalized Category Discovery (CVPR 2026 · better on 1 of 1 shared result)
- Happy: A Debiased Learning Framework for Continual Generalized Category Discovery (NeurIPS 2024 · better on 2 of 2 shared results)
- PromptCCD: Learning Gaussian Mixture Prompt Pool for Continual Category Discovery (ECCV 2024 · better on 1 of 1 shared result)
- Category Adaptation Meets Projected Distillation in Generalized Continual Category Discovery (ECCV 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
- Virtual Category-Guided Continual Generalized Category Discovery (ECCV 2026 · similarity 0.93)
- Happy: A Debiased Learning Framework for Continual Generalized Category Discovery (NeurIPS 2024 · similarity 0.93)
- Continual Generalized Category Discovery: Learning and Forgetting from a Bayesian Perspective (ICML 2025 · similarity 0.92)
- Solving the Catastrophic Forgetting Problem in Generalized Category Discovery (CVPR 2024 · similarity 0.91)
- Generalized Category Discovery via Reciprocal Learning and Class-Wise Distribution Regularization (ICML 2025 · similarity 0.88)
- When Domain Generalization meets Generalized Category Discovery: An Adaptive Task-Arithmetic Driven Approach (CVPR 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.
