研究人员正在探索先进的神经算子框架以增强科学计算。一篇论文介绍了无限阶核神经算子（IKNO），它使用无限阶核积分来提高表达能力，并在各种基准测试中达到了最先进的精度。另一项研究提出了一个统一的抽象神经流框架，证明了其在有限维函数逼近和无限维算子逼近方面的通用逼近性质，适用于神经网络和神经算子。 AI
影响 神经算子框架的这些进展可能为复杂的科学和工程问题带来更准确、更有效的解决方案。
排序理由 该集群包含多篇学术论文，详细介绍了神经算子的新理论框架和模型。
在 arXiv cs.NE (Neural & Evolutionary) 阅读 →
AI 生成摘要 · Google Gemini · 来自 5 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2603.01406v2 Announce Type: replace-cross Abstract: Neural PDE solvers are often described as learning solution operators that map problem data to PDE solutions. In this work, we argue that this interpretation is generally incorrect when boundary conditions vary. We show th…
arXiv:2605.22182v1 Announce Type: new Abstract: Neural operators have achieved significant success in modern scientific computing due to their flexibility and strong generalization capabilities. Existing models, however, primarily rely on first-order kernel integral approximation…
arXiv:2605.22557v1 Announce Type: new Abstract: We introduce an abstract neural flow framework for neural networks and neural operators. The framework contains two continuous-depth models, namely neural flows with composition and separation structures, and covers both finite-dime…
We introduce an abstract neural flow framework for neural networks and neural operators. The framework contains two continuous-depth models, namely neural flows with composition and separation structures, and covers both finite-dimensional function approximation and infinite-dime…
Neural operators have emerged as a powerful, discretization-invariant framework for solving partial differential equations (PDEs). Although established approaches like the Deep Operator Network (DeepONet) have successfully achieved universal approximation for operators, and archi…