Research map / Label-efficient and robust learning / Knowledge distillation
Knowledge Extraction with No Observable Data
Jaemin Yoo, Minyong Cho, Taebum Kim, 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 |
|---|---|---|---|
| fine-tune Tucker student with KEGNET-generated data | Tucker-compressed student · Pendigits | accuracy: 26.44% -> 82.62 ± 1.05% | Improved |
| fine-tune Tucker student with KEGNET-generated data | Tucker-compressed student · SVHN Student 2 | accuracy: 11.02% -> 87.26 ± 0.46% | Improved |
| fine-tune Tucker student with KEGNET-generated data | Tucker-compressed student · Fashion MNIST Student 3 | accuracy: 46.55% -> 79.95 ± 1.36% | Improved |
| use KEGNET data instead of Gaussian samples | T+Gaussian · MNIST Student 3 | accuracy: 71.76 ± 0.18% -> 89.94 ± 0.08% | Improved |
| use KEGNET data instead of uniform random samples | T+Uniform · Letter | accuracy: 62.50 ± 0.90% -> 77.73 ± 0.33% | Improved |
| fine-tune Tucker student with random Uniform/Gaussian data | Tucker-compressed student · Fashion MNIST | accuracy: < 65.09% for Student 1 | Hurt |
| fine-tune Tucker student with random Uniform/Gaussian data | Tucker-compressed student · MNIST Student 1 | accuracy: 85.18% -> 95.48 ± 0.11% | Improved |
| fine-tune Tucker student with random noise on complex datasets | Tucker-compressed student · SVHN Student 3 | accuracy: 11.07% -> 23.83 ± 1.86% | Mixed |
| add decoder loss and diversity loss to generator objective (ablation) | KEGNET generator · image datasets | data diversity | Improved |
Closest papers by meaning
- Small Scale Data-Free Knowledge Distillation (CVPR 2024 · similarity 0.89)
- Harmonizing knowledge Transfer in Neural Network with Unified Distillation (ECCV 2024 · similarity 0.89)
- Discovering and Overcoming Limitations of Noise-engineered Data-free Knowledge Distillation (NeurIPS 2022 · similarity 0.88)
- Dataset Distillation via Knowledge Distillation: Towards Efficient Self-Supervised Pre-training of Deep Networks (ICLR 2025 · similarity 0.87)
- All You Need in Knowledge Distillation Is a Tailored Coordinate System (AAAI 2025 · similarity 0.86)
- Hybrid Data-Free Knowledge Distillation (AAAI 2025 · 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.
