Research map / Architectures and efficiency / Pruning and efficient tuning
Distance-Based Regularisation of Deep Networks for Fine-Tuning
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 |
|---|---|---|---|
| Use MARS-PGM instead of the strongest baseline | ResNet-101 fine-tuning · Aircraft | accuracy: 60.86 -> 72.04 | Helped |
| Use MARS-PGM instead of the strongest baseline | ResNet-101 fine-tuning · Butterfly | accuracy: 77.91 -> 79.50 | Helped |
| Use MARS-PGM instead of the strongest baseline | ResNet-101 fine-tuning · Flowers | accuracy: 86.49 -> 87.42 | Helped |
| Use ℓ²-PGM instead of the strongest baseline | ResNet-101 fine-tuning · Pets | accuracy: 90.50 -> 90.47 | No clear effect |
| Use MARS-PGM instead of the strongest baseline | ResNet-101 fine-tuning · PubFig | accuracy: 84.38 -> 88.75 | Helped |
| Use ℓ²-PGM instead of the strongest baseline | ResNet-101 fine-tuning · DTD | accuracy: 69.87 -> 70.97 | Helped |
| Use MARS-PGM instead of the strongest baseline | ResNet-101 fine-tuning · Caltech | accuracy: 80.04 -> 79.11 | Hurt |
| Use MARS-PGM instead of the strongest baseline | EfficientNetB0 fine-tuning · Aircraft | accuracy: 70.61 -> 75.22 | Helped |
| Use MARS-PGM instead of the strongest baseline | EfficientNetB0 fine-tuning · Butterfly | accuracy: 79.61 -> 82.32 | Helped |
| Use MARS-PGM instead of ℓ²-PGM | EfficientNetB0 fine-tuning · Flowers | accuracy: 86.95 -> 90.36 | Helped |
| Use MARS-PGM instead of the strongest baseline | EfficientNetB0 fine-tuning · Pets | accuracy: 89.27 -> 91.38 | Helped |
| Use MARS-PGM instead of the strongest baseline | EfficientNetB0 fine-tuning · PubFig | accuracy: 88.62 -> 90.30 | Helped |
Closest papers by meaning
- Robust Fine-Tuning of Deep Neural Networks with Hessian-based Generalization Guarantees (ICML 2022 · similarity 0.92)
- Norm-Based Generalisation Bounds for Deep Multi-Class Convolutional Neural Networks (AAAI 2021 · similarity 0.89)
- A Theoretical Analysis of Fine-tuning with Linear Teachers (NeurIPS 2021 · similarity 0.88)
- Improved Regularization and Robustness for Fine-tuning in Neural Networks (NeurIPS 2021 · similarity 0.88)
- Growing a Brain: Fine-Tuning by Increasing Model Capacity (CVPR 2017 · similarity 0.87)
- Generalization of Scaled Deep ResNets in the Mean-Field Regime (ICLR 2024 · similarity 0.86)
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.