Research map / Label-efficient and robust learning / Recognition and long-tail
Combining Discrepancy-Confusion Uncertainty and Calibration Diversity for Active Fine-Grained Image Classification
Yinghao Jin, Xi Yang
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 uncertainty sampling (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
| use both discrepancy and confusion uncertainty rather than a single measure (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
| add calibration diversity after uncertainty-weighted clustering (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
| set local fusion size eta to 0.1 rather than 0.05 or lower (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
| set local fusion size eta to 0.2 or larger (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Hurt |
| use dynamic uncertainty threshold zeta over AL cycles instead of fixed zeta (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
| increase uncertainty strength lambda with larger annotation budget (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
| set diversity balance xi to 0.8 (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
| set diversity balance xi to 0.4 or lower at small budget (ablation) | DECERN · Caltech101 (ResNet50) | Top-1 accuracy | Hurt |
| use DECERN instead of K-Means, CoreSet or NoiseStability | Active learning baselines · Caltech101 (ResNet50) | Top-1 accuracy | Improved |
Closest papers by meaning
- ActiveDC: Distribution Calibration for Active Finetuning (CVPR 2024 · similarity 0.86)
- Cross-Layer and Cross-Sample Feature Optimization Network for Few-Shot Fine-Grained Image Classification (AAAI 2024 · similarity 0.85)
- Boundary Matters: A Bi-Level Active Finetuning Method (NeurIPS 2024 · similarity 0.85)
- Few-Shot Fine-Grained Image Classification with Progressively Feature Refinement and Continuous Relationship Modeling (AAAI 2025 · similarity 0.85)
- Fine-grained Classes and How to Find Them (ICML 2024 · similarity 0.85)
- Twofold Debiasing Enhances Fine-Grained Learning with Coarse Labels (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.
