Research map / Learning theory and algorithms / Kernels and regression theory
Toward Large Kernel Models
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 |
|---|---|---|---|
| k-means center selection instead of random centers (ablation) | EigenPro 3.0 · MNIST, CIFAR10, FashionMNIST (Table 3, p=100 to 10000) | accuracy (%): e.g., MNIST p=100: 91.89 vs 87.24; CIFAR10 p=10000: 52.72 vs 49.92 | Helped |
| scale model size (number of centers) with fixed data n (ablation) | EigenPro 3.0 general kernel models · Real data experiments (n=5M) | qualitative performance | Helped |
| increase number of training samples at fixed model size (ablation) | EigenPro 3.0 · Real data experiments | qualitative performance | Helped |
| data augmentation (MixUp, crop+flip, Gaussian noise) with original data as centers | EigenPro 3.0 kernel models · CIFAR10, MNIST, FashionMNIST | accuracy | Helped |
| Nystrom-based EigenPro 3.0 scaling to 1M centers vs FALKON limited to 256K centers | EigenPro 3.0 vs FALKON · Real data experiments | qualitative performance | Helped |
Closest papers by meaning
- Fast Training of Large Kernel Models with Delayed Projections (NeurIPS 2025 · similarity 0.91)
- Convolutional Deep Kernel Machines (ICLR 2024 · similarity 0.87)
- Stochastic Kernel Regularisation Improves Generalisation in Deep Kernel Machines (NeurIPS 2024 · similarity 0.87)
- Generalization in Kernel Regression Under Realistic Assumptions (ICML 2024 · similarity 0.87)
- Joker: Joint Optimization Framework for Lightweight Kernel Machines (ICML 2025 · similarity 0.86)
- KernelBench: Can LLMs Write Efficient GPU Kernels? (ICML 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.