Research map / Learning theory and algorithms / Matrix and tensor methods
Neural Spectral Decomposition for Dataset Distillation
Shaolei Yang, Shen Cheng, Mingbo Hong, Haoqiang Fan, Xing Wei, Shuaicheng Liu
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 |
|---|---|---|---|
| add spectral decomposition to MTT (ablation) | MTT · CIFAR10 IPC=1 | accuracy: 46.3 -> 67.9 | Improved |
| add real-distribution guided loss on top of decomposition (ablation) | MTT + decomposition · CIFAR10 IPC=1 | accuracy: 67.9 -> 68.5 | Improved |
| add real-distribution guided loss on top of decomposition (ablation) | MTT + decomposition · CIFAR10 IPC=10 | accuracy: 71.8 -> 73.4 | Improved |
| make DCT kernel learnable (ablation) | DCT decomposition · CIFAR10 IPC=1 | accuracy: 57.9 -> 62.3 | Improved |
| make SVD kernel learnable (ablation) | SVD decomposition · CIFAR10 IPC=1 | accuracy: 49.0 -> 53.8 | Improved |
| use learned kernel (ours) instead of fixed DCT (ablation) | NSD · CIFAR10 IPC=1 | accuracy: 57.9 -> 68.5 | Improved |
| plug decomposition into DM | DM · CIFAR10 IPC=10 (ConvNet) | accuracy: 48.9 -> 50.1 | Improved |
| plug decomposition into DC | DC · CIFAR10 IPC=10 (ConvNet) | accuracy: 44.9 -> 55.6 | Improved |
| plug decomposition into MTT, evaluate on AlexNet | MTT · CIFAR10 IPC=10 cross-arch | accuracy: 34.2 -> 71.4 | Improved |
| use moderate t3 size for spectrum tensor (ablation) | NSD · CIFAR10 IPC=1/10 | accuracy | Improved |
Closest papers by meaning
- Large Scale Dataset Distillation with Domain Shift (ICML 2024 · similarity 0.89)
- D^4: Dataset Distillation via Disentangled Diffusion Model (CVPR 2024 · similarity 0.87)
- Boost Self-Supervised Dataset Distillation via Parameterization, Predefined Augmentation, and Approximation (ICLR 2025 · similarity 0.86)
- Over-parameterized Student Model via Tensor Decomposition Boosted Knowledge Distillation (NeurIPS 2024 · similarity 0.86)
- Dataset Distillation for Pre-Trained Self-Supervised Vision Models (NeurIPS 2025 · similarity 0.86)
- Emphasizing Discriminative Features for Dataset Distillation in Complex Scenarios (CVPR 2025 · similarity 0.85)
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.
