Multifidelity data fusion in convolutional encoder/decoder networks
Multifidelity data fusion in convolutional encoder/decoder networks
复制标题
DOI：
10.1016/j.jcp.2022.111666
复制
发表时间：
2022-04
期刊：
影响因子：
--
通讯作者：
Lauren Partin;G. Geraci;A. Rushdi;M. Eldred;D. Schiavazzi
中科院分区：
文献类型：
--
作者：
Lauren Partin;G. Geraci;A. Rushdi;M. Eldred;D. Schiavazzi
我们分析了由编码器、解码器和跳跃连接组成的卷积神经网络的回归精度，并使用多保真度数据进行训练。除了需要比等效全连接网络少得多的可训练参数外，编码器、解码器、编码器-解码器或解码器-编码器架构可以学习任意维度的输入到输出之间的映射。我们在从一维函数到二维泊松方程求解器的模型中生成的一些高保真度和许多低保真度数据上进行训练时证明了它们的准确性。我们最后讨论了一些实现选择，以提高蒙特卡罗DropBlocks生成的不确定性估计的可靠性，并比较了低保真度、高保真度和多保真度方法的不确定性估计。
We analyze the regression accuracy of convolutional neural networks assembled from encoders, decoders and skip connections and trained with multifidelity data. Besides requiring significantly less trainable parameters than equivalent fully connected networks, encoder, decoder, encoder-decoder or decoder-encoder architectures can learn the mapping between inputs to outputs of arbitrary dimensionality. We demonstrate their accuracy when trained on a few high-fidelity and many low-fidelity data generated from models ranging from one-dimensional functions to Poisson equation solvers in two-dimensions. We finally discuss a number of implementation choices that improve the reliability of the uncertainty estimates generated by Monte Carlo DropBlocks, and compare uncertainty estimates among low-, high- and multifidelity approaches.