Research map / Architectures and efficiency / CNN design
Dataset Distillation with Infinitely Wide Convolutional Networks
Timothy Nguyen, Roman Novak, Lechao Xiao, Jaehoon Lee
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 |
|---|---|---|---|
| Use a convolutional NTK rather than the strongest listed prior distillation baseline | Dataset distillation · CIFAR-10, 1 image/class | test accuracy: 40.5±0.4 -> 64.7±0.2 | Improved |
| Use regularized ZCA preprocessing (ablation) | KIP ConvNet · CIFAR-10, 10 images/class | KIP accuracy: 61.0±0.2 -> 75.6±0.2 | Improved |
| Optimize support labels instead of keeping them fixed (ablation) | KIP ConvNet · CIFAR-10, 10 images/class, ZCA and no augmentation | KIP accuracy: 67.6±0.3 -> 75.6±0.2 | Improved |
| Add target-data augmentation (ablation) | KIP ConvNet · CIFAR-10, 10 images/class, ZCA and fixed labels | KIP accuracy: 67.6±0.3 -> 66.5±0.3 | Hurt |
| Add target-data augmentation (ablation) | KIP ConvNet · Fashion-MNIST, 10 images/class | KIP accuracy: 91.0±0.1 -> 88.8±0.1 | Hurt |
| Train with a family of kernels rather than a single kernel | Sampling KIP · CIFAR-10, 100 images | test performance across kernels | Improved |
| Transfer KIP-distilled data from KRR to finite-width neural network training | KIP ConvNet · CIFAR-10, 10 images/class | test accuracy: 53.1±0.5 -> 62.7±0.3 | Improved |
| Use label-solved data for finite-width NN training instead of KIP data | CIFAR-10 distilled dataset · CIFAR-10, 10 images/class | test accuracy: 62.7±0.3 -> 49.3±0.1 | Hurt |
| Use label-solved data for finite-width NN training instead of KIP data | SVHN distilled dataset · SVHN, 10 images/class | test accuracy: 75.0±0.1 -> 53.2±0.3 | Hurt |
Closest papers by meaning
- Large Scale Dataset Distillation with Domain Shift (ICML 2024 · similarity 0.90)
- Boost Self-Supervised Dataset Distillation via Parameterization, Predefined Augmentation, and Approximation (ICLR 2025 · similarity 0.89)
- Rethinking Dataset Distillation for Classification: Do Distilled Sets Outperform Coresets? (NeurIPS 2026 · similarity 0.89)
- Dataset Distillation for Pre-Trained Self-Supervised Vision Models (NeurIPS 2025 · similarity 0.88)
- Data Distillation Can Be Like Vodka: Distilling More Times For Better Quality (ICLR 2024 · similarity 0.88)
- Provable and Efficient Dataset Distillation for Kernel Ridge Regression (NeurIPS 2024 · similarity 0.88)
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.
