Research map / Label-efficient and robust learning / Recognition and long-tail
Transferability Estimation Using Bhattacharyya Class Separability
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use Gaussian Bhattacharyya class-overlap estimates to rank source architectures | GBC · Image classification source selection across 8 datasets | weighted Kendall τw: 0.39 -> 0.47 | Helped |
| Use Gaussian Bhattacharyya class-overlap estimates to rank target datasets | GBC · Image classification target selection | weighted Kendall τw: 0.82 | No clear effect |
| Use GBC rather than LogME or H-score for fixed-target source selection | Semantic segmentation transferability estimation · Semantic segmentation, fixed target | top-1 selection accuracy: 0.59 -> 0.65 | Helped |
| Use GBC rather than LEEP for fixed-source target selection | Semantic segmentation transferability estimation · Semantic segmentation, fixed source | weighted Kendall τw: 0.62 -> 0.69 | Helped |
| Add covariance regularization, with spherical regularization performing best (ablation) | GBC · CIFAR-10 source selection | weighted Kendall τw | Helped |
| Use 64-dimensional PCA rather than lower-dimensional projections (ablation) | GBC · CIFAR-10 source selection | weighted Kendall τw | Helped |
| Increase PCA projection beyond 64 dimensions (ablation) | GBC · CIFAR-10 source selection | weighted Kendall τw | Hurt |
| Use diagonal covariance rather than spherical covariance (ablation) | GBC · Semantic segmentation | transferability estimation performance | Helped |
Closest papers by meaning
- Transferability Metrics for Selecting Source Model Ensembles (CVPR 2022 · similarity 0.88)
- Implicit Modeling for Transferability Estimation of Vision Foundation Models (NeurIPS 2025 · similarity 0.86)
- Enhancing Generalized Few-Shot Semantic Segmentation via Effective Knowledge Transfer (AAAI 2025 · similarity 0.83)
- Understanding the Transferability of Representations via Task-Relatedness (NeurIPS 2024 · similarity 0.83)
- A High-Dimensional Statistical Method for Optimizing Transfer Quantities in Multi-Source Transfer Learning (NeurIPS 2025 · similarity 0.83)
- On Transferring Transferability: Towards a Theory for Size Generalization (NeurIPS 2025 · similarity 0.83)
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.