一项发表在arXiv上的新研究调查了神经网络代理模型在捕捉随机问题的完整解分布方面的性能，特别关注分布的尾部。研究人员发现，极端预测点的预测误差可能比平均场误差大一个数量级，这通常是由于网络在训练数据之外进行外推所致。该研究比较了前馈网络和深度算子网络，并提出使用弱形式残差损失训练的全连接神经网络在处理这些外插输入方面表现最佳。 AI
影响 强调了当前神经网络在预测极端结果方面的局限性，这对于物理建模中的风险评估至关重要。
排序理由 该集群包含一篇详细介绍神经网络性能数值研究的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
Neural network surrogate models have emerged as a promising approach to model solution fields for a wide variety of boundary value problems encountered in physical modeling. Stochastic problems represent an area of particularly high interest because of the potential to significan…
arXiv:2605.16078v1 Announce Type: new Abstract: Neural network surrogate models have emerged as a promising approach to model solution fields for a wide variety of boundary value problems encountered in physical modeling. Stochastic problems represent an area of particularly high…