Deep Operator Network
PulseAugur coverage of Deep Operator Network — every cluster mentioning Deep Operator Network across labs, papers, and developer communities, ranked by signal.
2 天有情绪数据
-
New SON framework quantifies uncertainty in SPDEs
Researchers have developed a new framework called the Stochastic Operator Network (SON) for quantifying uncertainty in stochastic partial differential equations (SPDEs). This method combines Deep Operator Networks with …
-
神经网络在不确定性传播的极端预测中遇到困难
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。
-
Deep Operator Networks predict composite material deformation with uncertainty quantification
Researchers have developed a Deep Operator Network (DeepONet) to predict process-induced deformation in carbon/epoxy composites. This data-driven surrogate model combines physics-based simulations with experimental meas…