研究发现：神经网络在组合任务上表现优于NTK限制

报道来源 [3]

1. Hugging Face Daily Papers TIER_1 English(EN) · 2026-07-07 15:21

   A Function-Space Dichotomy for Compositional Learning: Exponential Sub-Optimality of the Neural Tangent Kernel

   A persistent empirical observation is that trained neural networks outperform their neural tangent kernel (NTK) limit on tasks with compositional structure, yet a quantitative account of $\textbf{when}$ and $\textbf{by how much}$ has been lacking. Working on the unit circle, we g…

2. arXiv stat.ML TIER_1 English(EN) · Arkaprabha Ganguli, Emil Constantinescu · 2026-07-08 04:00

   组合学习的函数空间二分法：神经切线核的指数级次优性

   arXiv:2607.06382v1 Announce Type: new Abstract: A persistent empirical observation is that trained neural networks outperform their neural tangent kernel (NTK) limit on tasks with compositional structure, yet a quantitative account of $\textbf{when}$ and $\textbf{by how much}$ ha…

3. arXiv stat.ML TIER_1 English(EN) · Emil Constantinescu · 2026-07-07 15:21

   组合学习的函数空间二分法：神经切线核的指数级次优性

   A persistent empirical observation is that trained neural networks outperform their neural tangent kernel (NTK) limit on tasks with compositional structure, yet a quantitative account of $\textbf{when}$ and $\textbf{by how much}$ has been lacking. Working on the unit circle, we g…