Research map / Label-efficient and robust learning / Noisy labels
Improved Regularization and Robustness for Fine-tuning in Neural Networks
Read the paperCheck 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 |
|---|---|---|---|
| layer-wise regularization plus self-labeling (REGSL) vs L2-PGM | ResNet-101 fine-tuning · Stanford-Dogs | top-1 accuracy (%): 86.48±0.28 -> 89.58±0.13 | Helped |
| REGSL vs vanilla fine-tuning | ResNet-101 · Aircrafts | top-1 accuracy (%): 73.96±0.34 -> 75.32±0.23 | Helped |
| REGSL vs L2-SP on Stanford-Cars-like tasks | ResNet-101 · Stanford-Cars | top-1 accuracy (%): 89.28±0.14 -> 89.14±0.22 | No clear effect |
| REGSL vs ELR under label noise | ResNet-18 on Indoor · Indoor 80% independent noise | top-1 accuracy (%): 31.24±0.19 -> 34.08±0.79 | Helped |
| REGSL vs L2-PGM under label noise | ResNet-18 on Indoor · Indoor 20% independent noise | top-1 accuracy (%): 69.45±0.19 -> 72.51±0.46 | Helped |
| REGSL vs L2-PGM on few-shot | ResNet-12 · miniImageNet 5-way-5-shot | accuracy (%): 77.33 ± 0.59 -> 78.03 ± 0.54 | Helped |
| remove any one component (layer-wise reg, label correction, reweighting) (ablation) | REGSL · Indoor label noise | accuracy | Helped |
Closest papers by meaning
- Understanding and Mitigating the Label Noise in Pre-training on Downstream Tasks (ICLR 2024 · similarity 0.88)
- Tuning Stable Rank Shrinkage: Aiming at the Overlooked Structural Risk in Fine-tuning (CVPR 2024 · similarity 0.88)
- Understanding Overadaptation in Supervised Fine-Tuning: The Role of Ensemble Methods (ICML 2025 · similarity 0.87)
- Fine-Tuning is Fine, if Calibrated (NeurIPS 2024 · similarity 0.87)
- Domain-Aware Fine-Tuning: Enhancing Neural Network Adaptability (AAAI 2024 · similarity 0.87)
- TTE: Two Tokens Are Enough to Improve Parameter-Efficient Tuning (AAAI 2025 · similarity 0.87)
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.