Research map / Label-efficient and robust learning / Knowledge distillation
Factorizing Knowledge in Neural Networks
Xingyi Yang, Jingwen Ye, Xinchao Wang
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 |
|---|---|---|---|
| factorize teacher into CKN plus task-specific network (KF) instead of baseline MBNv2 | MBNv2 · CIFAR-10 | accuracy: 93.58 -> 94.80 | Improved |
| use KF (ResNet-18 CKN + MBNv2x0.5 TSN) instead of WRN-28-10 baseline | WRN-28-10 · CIFAR-10 | accuracy: 95.32 -> 95.40 | Improved |
| use KF (WRN-28-2 CKN + 2 TSN) instead of ResNet-18 baseline | ResNet-18 · CIFAR-10 | accuracy: 94.45 -> 95.03 | Improved |
| add knowledge distillation to MBNv2 on CIFAR-10 sub-tasks | MBNv2 · CIFAR-10 binary and trinary sub-tasks | accuracy: +2% | Improved |
| replace distillation with knowledge factorization on sub-tasks | MBNv2 · CIFAR-10 binary and trinary sub-tasks | accuracy: +1% | Improved |
| use factor networks instead of KD/baseline under adversarial attack | MBNv2, ResNet-18 · CIFAR-10 FGSM and PGD-40 | robust accuracy | Improved |
Closest papers by meaning
- Toward a Unified Statistical Theory of Unsupervised Pretraining and Supervised Neural Knowledge Graph Learning (NeurIPS 2026 · similarity 0.82)
- Factor Graph-based Interpretable Neural Networks (ICLR 2025 · similarity 0.81)
- Explainable Neural Networks with Guarantee: A Sparse Estimation Approach (AAAI 2025 · similarity 0.81)
- Factorized Explainer for Graph Neural Networks (AAAI 2024 · similarity 0.81)
- Defining and extracting generalizable interaction primitives from DNNs (ICLR 2024 · similarity 0.80)
- Layerwise Change of Knowledge in Neural Networks (ICML 2024 · similarity 0.80)
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.
