Comparison of Deep Neural Networks and Deep Hierarchical Models for Spatio-Temporal Data
Comparison of Deep Neural Networks and Deep Hierarchical Models for Spatio-Temporal Data
复制标题
DOI：
10.1007/s13253-019-00361-7
复制
发表时间：
2019-02
期刊：
影响因子：
--
通讯作者：
C. Wikle
中科院分区：
文献类型：
--
作者：
C. Wikle
时空数据在农业、生态和环境科学中无处不在，它们的研究对于理解和预测各种各样的过程非常重要。对随时间变化的空间过程建模的困难之一是必须描述这种过程如何变化的依赖结构的复杂性，以及高维复杂数据集和大型预测域的存在。为非线性动态时空模型（DSTMs）指定参数化是一项特别具有挑战性的工作，它同时具有科学意义和计算效率。统计学家已经开发了多层次（深度）层次模型，可以适应过程的复杂性以及预测和推理中的不确定性。然而，这些模型可能是昂贵的，并且通常是特定于应用程序的。另一方面，机器学习社区已经为非线性时空建模开发了替代的“深度学习”方法。这些模型是灵活的，但通常没有在概率框架中实现。这两种范式有许多共同之处，并提出了可以从每个框架的元素中受益的混合方法。本文简要介绍了多层次（深度）分层DSTM （H-DSTM）框架和机器学习中的深度模型，最后介绍了深度神经DSTM （DN-DSTM）。结合h - dstm和回波状态网络dn - dstm元素的最新方法作为例证。本文附带的补充资料出现在网上。
Spatio-temporal data are ubiquitous in the agricultural, ecological, and environmental sciences, and their study is important for understanding and predicting a wide variety of processes. One of the difficulties with modeling spatial processes that change in time is the complexity of the dependence structures that must describe how such a process varies, and the presence of high-dimensional complex datasets and large prediction domains. It is particularly challenging to specify parameterizations for nonlinear dynamic spatio-temporal models (DSTMs) that are simultaneously useful scientifically and efficient computationally. Statisticians have developed multi-level (deep) hierarchical models that can accommodate process complexity as well as the uncertainties in the predictions and inference. However, these models can be expensive and are typically application specific. On the other hand, the machine learning community has developed alternative “deep learning” approaches for nonlinear spatio-temporal modeling. These models are flexible yet are typically not implemented in a probabilistic framework. The two paradigms have many things in common and suggest hybrid approaches that can benefit from elements of each framework. This overview paper presents a brief introduction to the multi-level (deep) hierarchical DSTM (H-DSTM) framework, and deep models in machine learning, culminating with the deep neural DSTM (DN-DSTM). Recent approaches that combine elements from H-DSTMs and echo state network DN-DSTMs are presented as illustrations. Supplementary materials accompanying this paper appear online.