Research map / Architectures and efficiency / Pruning and efficient tuning
Data 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 |
|---|---|---|---|
| Apply a learned universal perturbation to the input data | VGGFace + NNET · LFW, Gender (intra-dataset) | accuracy (%): 87.94 -> 91.17 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · CelebA, Gender (intra-dataset) | accuracy (%): 82.13 -> 83.08 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · MUCT, Gender (intra-dataset) | accuracy (%): 91.67 -> 94.31 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · LFW, Smiling (intra-dataset) | accuracy (%): 76.18 -> 82.42 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · LFW, Bushy Eyebrows (intra-dataset) | accuracy (%): 68.34 -> 69.98 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · LFW, Pale Skin (intra-dataset) | accuracy (%): 72.83 -> 74.81 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · CelebA, Smiling (intra-dataset) | accuracy (%): 67.82 -> 71.30 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · CelebA, Attractive (intra-dataset) | accuracy (%): 70.48 -> 70.54 | Helped |
| Apply a learned universal perturbation to the input data | VGGFace + NNET · CelebA, Wearing Lipstick (intra-dataset) | accuracy (%): 80.95 -> 81.29 | Helped |
| Apply a learned universal perturbation to a dataset different from the model's training data | Gender classifier · MUCT using a model trained on LFW | accuracy (%): 57.84 -> 83.65 | Helped |
| Apply a learned universal perturbation to a dataset different from the model's training data | Gender classifier · LFW using a model trained on CelebA | accuracy (%): 56.01 -> 86.33 | Helped |
| Apply a learned universal perturbation to a dataset different from the model's training data | Attribute classifiers · LFW, Bushy Eyebrows using a CelebA-trained model | accuracy (%): 45.40 -> 68.91 | Helped |
Closest papers by meaning
- Fine-Tuning is Fine, if Calibrated (NeurIPS 2024 · similarity 0.85)
- Tuning Stable Rank Shrinkage: Aiming at the Overlooked Structural Risk in Fine-tuning (CVPR 2024 · similarity 0.82)
- TTE: Two Tokens Are Enough to Improve Parameter-Efficient Tuning (AAAI 2025 · similarity 0.82)
- Mitigating Feature Gap for Adversarial Robustness by Feature Disentanglement (AAAI 2025 · similarity 0.81)
- Quick-Tune: Quickly Learning Which Pretrained Model to Finetune and How (ICLR 2024 · similarity 0.81)
- Recovering the Pre-Fine-Tuning Weights of Generative Models (ICML 2024 · similarity 0.81)
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.