Continuous-Time Stochastic Differential Networks for Irregular Time Series Modeling
Continuous-Time Stochastic Differential Networks for Irregular Time Series Modeling
复制标题
用于不规则时间序列建模的连续时间随机微分网络
DOI：
10.1007/978-3-030-92307-5_40
复制
发表时间：
2021
期刊：
影响因子：
--
通讯作者：
Jacqueline Wu
中科院分区：
文献类型：
--
作者：
Yingru Liu;Yucheng Xing;Xuewen Yang;Xin Wang;Jing Shi;Di Jin;Zhaoyue Chen;Jacqueline Wu
连续时间随机动力学学习是不规则时间序列建模的基础和关键问题，不规则时间序列的观测值在时间和维度上都是不规则和稀疏的。对于一个给定的系统，其潜在状态和观测数据是多元的，通常不可能推导出一个精确的连续时间随机过程来描述系统行为。为了解决上述问题，我们应用变分贝叶斯方法，提出了一种灵活的连续时间随机递归神经网络——变分随机微分网络（VSDN），该网络通过神经随机微分方程（SDE）嵌入不规则时间序列的复杂动力学。vsdn通过深度神经网络捕获潜在状态和观测值之间的随机依赖关系。我们还结合了两个差分证据下限来有效地训练模型。通过综合实验，我们表明vsdn优于最先进的连续时间深度学习模型，并在不规则时间序列的预测和插值任务上取得了显着的性能。
Learning continuous-time stochastic dynamics is a fundamental and essential problem in modeling irregular time series, whose observations are irregular and sparse in both time and dimension. For a given system whose latent states and observed data are multivariate, it is generally impossible to derive a precise continuous-time stochastic process to describe the system behaviors. To solve the above problem, we apply Variational Bayesian method and propose a flexible continuous-time stochastic recurrent neural network namedVariational Stochastic Differential Networks (VSDN), which embeds the complicated dynamics of the irregular time series by neural Stochastic Differential Equations (SDE). VSDNs capture the stochastic dependency among latent states and observations by deep neural networks. We also incorporate two differential Evidence Lower Bounds to efficiently train the models. Through comprehensive experiments, we show that VSDNs outperform state-of-the-art continuous-time deep learning models and achieve remarkable performance on prediction and interpolation tasks for irregular time series.