arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2007.08729math.NAcs.NA
收藏

高维近似的深度ReLU神经网络

Deep ReLU neural networks in high-dimensional approximation

Dinh Dũng, Van Kien Nguyen

摘要:

我们研究了深度ReLU(Rectified Linear Unit)神经网络在$d$维单位立方体上定义的混合光滑的Hölder-Zygmund空间中逼近函数的计算复杂性,当维度$d$可能非常大时。逼近误差以各向同性索伯列夫空间的范数度量。对于来自混合光滑的Hölder-Zygmund空间的每个函数$f$,我们显式地构建了一个深度ReLU神经网络,其输出以规定的精度$\varepsilon$近似$f$,并证明了这种近似的计算复杂性的严格的维度依赖的上限和下界,其特征在于这种深度ReLU神经网络的大小和深度,显式地在$d$和$\varepsilon$中。这些结果的证明特别依赖于基于Faber级数的稀疏网格采样恢复的近似。

英文摘要:

We study the computation complexity of deep ReLU (Rectified Linear Unit) neural networks for the approximation of functions from the Hölder-Zygmund space of mixed smoothness defined on the $d$-dimensional unit cube when the dimension $d$ may be very large. The approximation error is measured in the norm of isotropic Sobolev space. For every function $f$ from the Hölder-Zygmund space of mixed smoothness, we explicitly construct a deep ReLU neural network having an output that approximates $f$ with a prescribed accuracy $\varepsilon$, and prove tight dimension-dependent upper and lower bounds of the computation complexity of this approximation, characterized as the size and the depth of this deep ReLU neural network, explicitly in $d$ and $\varepsilon$. The proof of these results are in particular, relied on the approximation by sparse-grid sampling recovery based on the Faber series.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑