Research map / Architectures and efficiency / Pruning and efficient tuning
Exploring Model Transferability through the Lens of Potential Energy
Xiaotong Li, Zixuan Hu, Yixiao Ge, Ying Shan, Ling-Yu Duan
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 potential-energy-decline dynamic modeling | LogME · Cifar100 | weighted Kendall tau_w: -0.008 -> 0.667 | Improved |
| add potential-energy-decline dynamic modeling | GBC · Cifar100 | weighted Kendall tau_w: -0.249 -> 0.359 | Improved |
| add potential-energy-decline dynamic modeling | SFDA · Aircraft | weighted Kendall tau_w: 0.254 -> 0.464 | Improved |
| add potential-energy-decline dynamic modeling | SFDA · DTD | weighted Kendall tau_w: 0.749 -> 0.907 | Improved |
| add potential-energy-decline dynamic modeling | LogME · Aircraft | weighted Kendall tau_w: 0.223 -> 0.509 | Improved |
| add potential-energy-decline dynamic modeling | SFDA · Pets | weighted Kendall tau_w: 0.586 -> 0.462 | Hurt |
| add potential-energy-decline dynamic modeling | SFDA · Food | weighted Kendall tau_w: 0.685 -> 0.581 | Hurt |
| add potential-energy-decline dynamic modeling | GBC · Cifar10 | weighted Kendall tau_w: 0.008 -> 0.017 | No clear effect |
| add potential-energy-decline dynamic modeling | GBC · VOC | weighted Kendall tau_w: -0.041 -> -0.035 | No clear effect |
| grid-search elasticity coefficient k instead of default k=1.0 (ablation) | LogME+Ours · Caltech101 | weighted Kendall tau_w: 0.505 -> 0.611 | Improved |
| grid-search elasticity coefficient k instead of default k=1.0 (ablation) | SFDA+Ours · Cifar100 | weighted Kendall tau_w: 0.568 -> 0.744 | Improved |
| grid-search elasticity coefficient k instead of default k=1.0 (ablation) | GBC+Ours · Aircraft | weighted Kendall tau_w: 0.462 -> 0.549 | Improved |
Closest papers by meaning
- Implicit Modeling for Transferability Estimation of Vision Foundation Models (NeurIPS 2025 · similarity 0.87)
- Features are fate: a theory of transfer learning in high-dimensional regression (ICML 2025 · similarity 0.84)
- LEAD: Exploring Logit Space Evolution for Model Selection (CVPR 2024 · similarity 0.84)
- Learning Physics From Video: Unsupervised Physical Parameter Estimation for Continuous Dynamical Systems (CVPR 2025 · similarity 0.84)
- Energy-guided Entropic Neural Optimal Transport (ICLR 2024 · similarity 0.84)
- Langevin-Informed Transfer Learning: Replacing the Target Samples by Black-Box Feedback (NeurIPS 2026 · 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.
