1
The Neural Tangent Kernel for Classification
深入探讨神经正切核在分类任务中的理论分析，为神经网络泛化研究提供新视角。
arXiv:2605.17606v1 Announce Type: new Abstract: In wide neural networks, the Neural Tangent Kernel (NTK) remains approximately constant during trainin…
2
Mirror Mean-Field Langevin Dynamics
平均场朗之万动力学的镜像变体，为无穷宽神经网络训练提供新理论视角
arXiv:2505.02621v2 Announce Type: replace Abstract: The mean-field Langevin dynamics (MFLD) minimizes an entropy-regularized nonlinear convex function…