Saved in:
Bibliographic Details
Main Authors: Liu, Yirui, Qiao, Xinghao, Pei, Yulong, Wang, Liying
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2305.14543
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918295655940096
author Liu, Yirui
Qiao, Xinghao
Pei, Yulong
Wang, Liying
author_facet Liu, Yirui
Qiao, Xinghao
Pei, Yulong
Wang, Liying
contents This paper introduces the Deep Functional Factor Model (DF2M), a Bayesian nonparametric model designed for analysis of high-dimensional functional time series. DF2M is built upon the Indian Buffet Process and the multi-task Gaussian Process, incorporating a deep kernel function that captures non-Markovian and nonlinear temporal dynamics. Unlike many black-box deep learning models, DF2M offers an explainable approach to utilizing neural networks by constructing a factor model and integrating deep neural networks within the kernel function. Additionally, we develop a computationally efficient variational inference algorithm to infer DF2M. Empirical results from four real-world datasets demonstrate that DF2M provides better explainability and superior predictive accuracy compared to conventional deep learning models for high-dimensional functional time series.
format Preprint
id arxiv_https___arxiv_org_abs_2305_14543
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Deep Functional Factor Models: Forecasting High-Dimensional Functional Time Series via Bayesian Nonparametric Factorization
Liu, Yirui
Qiao, Xinghao
Pei, Yulong
Wang, Liying
Machine Learning
This paper introduces the Deep Functional Factor Model (DF2M), a Bayesian nonparametric model designed for analysis of high-dimensional functional time series. DF2M is built upon the Indian Buffet Process and the multi-task Gaussian Process, incorporating a deep kernel function that captures non-Markovian and nonlinear temporal dynamics. Unlike many black-box deep learning models, DF2M offers an explainable approach to utilizing neural networks by constructing a factor model and integrating deep neural networks within the kernel function. Additionally, we develop a computationally efficient variational inference algorithm to infer DF2M. Empirical results from four real-world datasets demonstrate that DF2M provides better explainability and superior predictive accuracy compared to conventional deep learning models for high-dimensional functional time series.
title Deep Functional Factor Models: Forecasting High-Dimensional Functional Time Series via Bayesian Nonparametric Factorization
topic Machine Learning
url https://arxiv.org/abs/2305.14543