Research map / Label-efficient and robust learning / Recognition and long-tail
Generate, Refine, and Encode: Leveraging Synthesized Novel Samples for On-the-Fly Fine-Grained Category Discovery
Xiao Liu, Nan Pu, Haiyang Zheng, 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 leader-based contrastive loss (SLE) (ablation) | SMILE + DiffGRE · CUB (hash-like, ablation) | ACC-ALL: 33.2 -> 35.4 | Improved |
| add leader-based contrastive loss (SLE) (ablation) | SMILE + DiffGRE · Arachnida (hash-like, ablation) | ACC-ALL: 29.3 -> 35.4 | Improved |
| add diversity-driven refinement (DDR) (ablation) | SMILE + DiffGRE · CUB (hash-like, ablation) | ACC-ALL: 32.4 -> 35.4 | Improved |
| add diversity-driven refinement (DDR) (ablation) | SMILE + DiffGRE · Arachnida (hash-like, ablation) | ACC-ALL: 33.5 -> 35.4 | Improved |
| add classification loss L_c (ablation) | SMILE + DiffGRE · CUB (hash-like, ablation) | ACC-ALL: 33.6 -> 35.4 | Improved |
| add synthesized samples without DDR (ablation) | SMILE · CUB (hash-like, ablation) | ACC-ALL: 32.2 -> 32.4 | Improved |
| add DiffGRE to BaseHash | BaseHash · OCD hash-like inference (CUB/Cars/Animalia avg) | ACC-ALL: 22.6 -> 32.5 | Improved |
| add DiffGRE to SMILE | SMILE · OCD hash-like inference (Arachnida/Mollusca/Pets avg) | ACC-ALL: 34.2 -> 38.2 | Improved |
| add DiffGRE to PHE | PHE · OCD online clustering inference (Arachnida/Mollusca/Pets avg) | ACC-ALL: 43.8 -> 46.6 | Improved |
| replace SLC with SLE-based online clustering | BaseHash · OCD online clustering inference (Arachnida/Mollusca/Pets avg) | ACC-ALL: 30.7 -> 42.7 | Improved |
| replace ACG with CutMix (ablation) | SMILE + DiffGRE · CUB (synthesis method comparison) | ACC-ALL: 35.4 -> 32.8 | Hurt |
| replace ACG with Diff-Mix on known categories (ablation) | SMILE + DiffGRE · CUB (synthesis method comparison) | ACC-ALL: 35.4 -> 32.9 | Hurt |
Closest papers by meaning
- TALON: Test-time Adaptive Learning for On-the-Fly Category Discovery (CVPR 2026 · similarity 0.89)
- On-the-Fly Category Discovery (CVPR 2023 · similarity 0.89)
- Prototypical Hash Encoding for On-the-Fly Fine-Grained Category Discovery (NeurIPS 2024 · similarity 0.88)
- Adaptive Gaussian Expansion for On-the-fly Category Discovery (ICLR 2026 · similarity 0.88)
- Prior-Constrained Association Learning for Fine-Grained Generalized Category Discovery (AAAI 2025 · similarity 0.86)
- Transfer and Alignment Network for Generalized Category Discovery (AAAI 2024 · similarity 0.86)
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.
