Research map / Label-efficient and robust learning / Recognition and long-tail
On-the-Fly Category Discovery
Ruoyi Du, Dongliang Chang, Kongming Liang, Timothy Hospedales, Yi-Zhe Song, Zhanyu Ma
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 |
|---|---|---|---|
| replace the baseline hash projection head with separate sign and magnitude branches for contrastive training, using the sign branch for inference (ablation) | Hash coding-based OCD model · CIFAR100 and CUB-200-2011 | All accuracy: 37.82 -> 43.94; 21.09 -> 24.93 | Improved |
| add magnitude regularization (ablation) | Hash coding-based OCD model · CIFAR100 and CUB-200-2011 | All accuracy: 37.82 -> 42.05; 21.09 -> 21.48 | Improved |
| combine sign-magnitude disentanglement with magnitude regularization (ablation) | Hash coding-based OCD model · CIFAR100 and CUB-200-2011 | All accuracy: 37.82 -> 51.59; 21.09 -> 32.24 | Improved |
| use SMILE rather than the strongest listed baseline | OCD · CIFAR10, CIFAR100, ImageNet-100, CUB-200-2011, Stanford Cars, HERB19 | All accuracy | Improved |
| vary the magnitude-regularization weight α (ablation) | SMILE · CIFAR100 and CUB-200-2011 | All accuracy: 49.96; 51.24; 51.59; 45.80; 42.69; 31.61; 32.03; 32.24; 30.14; 28.60 | Mixed |
| vary the smoothness parameter τ (ablation) | SMILE · CIFAR100 and CUB-200-2011 | All accuracy: 50.81; 50.81; 51.59; 49.16; 46.31; 31.43; 31.90; 32.24; 26.71; 19.35 | Mixed |
Closest papers by meaning
- Prototypical Hash Encoding for On-the-Fly Fine-Grained Category Discovery (NeurIPS 2024 · similarity 0.90)
- TALON: Test-time Adaptive Learning for On-the-Fly Category Discovery (CVPR 2026 · similarity 0.90)
- Assignment-Driven Hash Learning in a Hyper-Semantic Space for On-the-Fly Category Discovery (CVPR 2026 · similarity 0.89)
- Generate, Refine, and Encode: Leveraging Synthesized Novel Samples for On-the-Fly Fine-Grained Category Discovery (ICCV 2025 · similarity 0.89)
- Semantic-Guided Novel Category Discovery (AAAI 2024 · similarity 0.87)
- SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning (ICLR 2024 · 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.
