Research map / Continual learning / Class-incremental
DP-BOA: Dirichlet-Process Birth-or-Assign for On-the-Fly Category Discovery
Peiyan Gu, Zixin Teng, Xuming He
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 |
|---|---|---|---|
| freeze NIW posteriors vs online evidence-adaptive updates (ablation) | DP-BOA · CUB | Novel accuracy: 27.0 -> 51.6 | Improved |
| use evidence-adaptive posterior updates (ablation) | DP-BOA · Animalia | All accuracy: 40.2 -> 50.7 | Improved |
| use full covariance instead of spherical (ablation) | DP-BOA · CUB | All accuracy: 40.4 -> 53.4 | Improved |
| estimate Psi0 from support-set pooled covariance (ablation) | DP-BOA · CUB | All accuracy: 40.8 -> 53.4 | Improved |
| estimate kappa0 by empirical Bayes (ablation) | DP-BOA · Animalia | All accuracy: 45.2 -> 50.7 | Improved |
| add DP category prior (ablation) | DP-BOA · Animalia | Novel accuracy: 38.4 -> 43.7 | Improved |
| set prior mean mu0 to global mean (ablation) | DP-BOA · Animalia/CUB/OxfordPets | All accuracy: 50.1 -> 50.7 (Animalia) | No clear effect |
| replace full covariance with low-rank (DP-BOA-L) (ablation) | DP-BOA · CUB | All accuracy: 53.4 -> 52.2 | Mixed |
| DP-BOA head vs PHE under bursty streams | DP-BOA · CUB bursty stream | All accuracy: 33.8 -> 50.5 | Improved |
| DP-BOA vs PHE without external knowledge | DP-BOA · ImageNet100 | All accuracy: 34.0 -> 33.8 | No clear effect |
| DP-BOA vs PHE | DP-BOA · OxfordPets | All accuracy: 48.3 -> 59.0 | Improved |
| vary DP concentration log alpha from -6 to -12 (ablation) | DP-BOA · Animalia/CUB/OxfordPets | All accuracy: 48.8-51.2 (Animalia) | No clear effect |
Closest papers by meaning
- TALON: Test-time Adaptive Learning for On-the-Fly Category Discovery (CVPR 2026 · similarity 0.87)
- Generate, Refine, and Encode: Leveraging Synthesized Novel Samples for On-the-Fly Fine-Grained Category Discovery (ICCV 2025 · similarity 0.87)
- Unleashing the Potential of Model Bias for Generalized Category Discovery (AAAI 2025 · similarity 0.86)
- Continual Generalized Category Discovery: Learning and Forgetting from a Bayesian Perspective (ICML 2025 · similarity 0.85)
- Prototypical Hash Encoding for On-the-Fly Fine-Grained Category Discovery (NeurIPS 2024 · similarity 0.84)
- Prior-Constrained Association Learning for Fine-Grained Generalized Category Discovery (AAAI 2025 · similarity 0.84)
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.
