Abstract:In the last few decades, building regression models for non-scalar variables, including time series, text, image, and video, has attracted increasing interests of researchers from the data analytic community. In this paper, we focus on a multivariate time series regression problem. Specifically, we aim to learn mathematical mappings from multiple chronologically measured numerical variables within a certain time interval S to multiple numerical variables of interest over time interval T. Prior arts, including the multivariate regression model, the Seq2Seq model, and the functional linear models, suffer from several limitations. The first two types of models can only handle regularly observed time series. Besides, the conventional multivariate regression models tend to be biased and inefficient, as they are incapable of encoding the temporal dependencies among observations from the same time series. The sequential learning models explicitly use the same set of parameters along time, which has negative impacts on accuracy. The function-on-function linear model in functional data analysis (a branch of statistics) is insufficient to capture complex correlations among the considered time series and suffer from underfitting easily. In this paper, we propose a general functional mapping that embraces the function-on-function linear model as a special case. We then propose a non-linear function-on-function model using the fully connected neural network to learn the mapping from data, which addresses the aforementioned concerns in the existing approaches. For the proposed model, we describe in detail the corresponding numerical implementation procedures. The effectiveness of the proposed model is demonstrated through the application to two real-world problems.

Determining the Input Dimension of A Neural Network for Nonlinear Time Series Prediction

Determining the Input Time Delay Τ of a Neural Network for Nonlinear Time Series Prediction

Neural Network Emulation of Spatio-temporal Data Using Linear and Nonlinear Dimensionality Reduction

Determining the Minimum Embedding Dimension of Nonlinear Time Series Based on Prediction Method

Correlation Dimension Analysis: A Non-linear Time Series Analysis for Data Processing

Optical characterization of chevron texture formation in nematic electroconvection

Long Input Sequence Network for Long Time Series Forecasting

An enhanced hybrid method for time series prediction using linear and neural network models

Prediction of Time Series Data with Low Latitude Features.

Efficient Interpretable Nonlinear Modeling for Multiple Time Series

Adaptive Graph Convolutional Network Framework for Multidimensional Time Series Prediction

Adaptive deep learning for nonlinear time series models

Input Factor Selection Based on Interpretable Neural Network for Industrial Virtual Sensing Application

DDTCN: Decomposed dimension time-domain convolutional neural network along spatial dimensions for multiple long-term series forecasting

On the modelling and prediction of high-dimensional functional time series

A Method for Estimating the Entropy of Time Series Using Artificial Neural Networks

Deep Nonlinear Sufficient Dimension Reduction

A Non-linear Function-on-Function Model for Regression with Time Series Data

Time series prediction based on NARX neural networks: An advanced approach

Sparse-Input Neural Networks for High-dimensional Nonparametric Regression and Classification

Dirichlet Graph Convolution Coupled Neural Differential Equation for Spatio-temporal Time Series Prediction