Research map / Label-efficient and robust learning / Semi-supervised learning
Memory Consistency Guided Divide-and-Conquer Learning for Generalized Category Discovery
Yuanpeng Tu, Zhun Zhong, Yuxi Li, Hengshuang Zhao
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 MCDL to SimGCD | SimGCD · CUB | accuracy: +8.4 | Improved |
| add MCDL to SimGCD | SimGCD · Stanford Cars | accuracy: +8.1 | Improved |
| add MCDL to SimGCD | SimGCD · FGVC-Aircraft | accuracy: +3.4 | Improved |
| add MCDL to SimGCD | SimGCD · CIFAR100 | accuracy: 80.1 -> 84.0 | Improved |
| add MCDL to SimGCD | SimGCD · CIFAR10 | accuracy: 97.1 -> 97.3 | No clear effect |
| add MCDL to SimGCD | SimGCD · Herbarium19 (vanilla ACC) | accuracy: 44.0 -> 49.4 | Improved |
| add MCDL to GCD | GCD · CUB | accuracy: 51.3 -> 55.6 | Improved |
| add MCDL to CiPR | CiPR · CUB | accuracy: 57.1 -> 59.7 | Improved |
| add weak-augmentation memory bank M_w (ablation) | SimGCD · CUB | accuracy: 60.3 -> 66.8 | Improved |
| add strong-augmentation memory bank M_s alone (ablation) | SimGCD · CUB | accuracy: 60.3 -> 62.8 | Improved |
| add M_s on top of M_w (ablation) | SimGCD+MCDL · CUB | accuracy: 66.8 -> 68.7 | Improved |
| add L_sup (ablation) | SimGCD · CUB | accuracy: 60.3 -> 67.2 | Improved |
Closest papers by meaning
- Consistent Supervised-Unsupervised Alignment for Generalized Category Discovery (NeurIPS 2025 · similarity 0.89)
- Solving the Catastrophic Forgetting Problem in Generalized Category Discovery (CVPR 2024 · similarity 0.89)
- Generalized Category Discovery via Reciprocal Learning and Class-Wise Distribution Regularization (ICML 2025 · similarity 0.89)
- DebGCD: Debiased Learning with Distribution Guidance for Generalized Category Discovery (ICLR 2025 · similarity 0.89)
- Happy: A Debiased Learning Framework for Continual Generalized Category Discovery (NeurIPS 2024 · similarity 0.88)
- SEAL: Semantic-Aware Hierarchical Learning for Generalized Category Discovery (NeurIPS 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.
