Research map / Label-efficient and robust learning / Recognition and long-tail
Fast and Accurate Transferability Measurement by Evaluating Intra-class Feature Variance
Huiwen Xu, U Kang
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 |
|---|---|---|---|
| measure transferability by intra-class feature variance (TMI) instead of discriminability scores | transferability measurement, supervised models · 17 datasets, Kendall correlation | Kendall correlation: best in 13 of 17 datasets | Mixed |
| use TMI for top-5 architecture selection rather than TransRate | transferability measurement · 17 target tasks | top-5 hit ratio: 9 / 17 -> 14 / 17 | Improved |
| use TMI for best source data selection rather than TransRate | transferability measurement · ResNet-50 source data selection | top-1 hit ratio: 13 / 17 -> 17 / 17 | Improved |
| use conditional entropy as the intra-class variance measure instead of Contrast, Center, SNCA or MS terms (ablation) | TMI · 17 datasets, Kendall correlation | Kendall correlation: best in 15 of 17 datasets | Improved |
| apply TMI to self-supervised pre-trained models without classifiers | TMI · 11 self-supervised models, 17 datasets | Kendall correlation: best in 13 of 17; VisDA -0.597 | Mixed |
| apply TMI to transferring layer selection | TMI · ResNet-18, 17 datasets | Kendall correlation: best in 16 of 17 datasets | Improved |
| use TMI vs baselines to check consistency of transferability sign with accuracy gain from pre-training | TMI · ImageNet-pretrained vs from-scratch ResNet-50 | sign consistency: 14 of 17 cases consistent | Improved |
| vary number of neighbors n_k in the entropy estimator (ablation) | TMI · hyperparameter sensitivity | correlation | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- LogMe — LogME: Practical Assessment of Pre-trained Models for Transfer Learning (ICML 2021) · better on 0 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Frustratingly Easy Transferability Estimation (ICML 2022 · similarity 0.90)
- Understanding the Transferability of Representations via Task-Relatedness (NeurIPS 2024 · similarity 0.88)
- A Mathematical Framework for Quantifying Transferability in Multi-source Transfer Learning (NeurIPS 2021 · similarity 0.87)
- Implicit Modeling for Transferability Estimation of Vision Foundation Models (NeurIPS 2025 · similarity 0.87)
- Transferring Knowledge From Large Foundation Models to Small Downstream Models (ICML 2024 · similarity 0.85)
- Features are fate: a theory of transfer learning in high-dimensional regression (ICML 2025 · 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.
