Deep Factors with Gaussian Processes for Forecasting

A large collection of time series poses significant challenges for classical and neural forecasting approaches. Classical time series models fail to fit data well and to scale to large problems, but succeed at providing uncertainty estimates. The converse is true for deep neural networks. In this pa...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2018-11
Main Authors: Maddix, Danielle C, Wang, Yuyang, Smola, Alex
Format: Article
Language:English
Subjects:
Forecasting
Gaussian process
Materials handling
Mathematical models
Neural networks
State of the art
Time series
Uncertainty
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A large collection of time series poses significant challenges for classical and neural forecasting approaches. Classical time series models fail to fit data well and to scale to large problems, but succeed at providing uncertainty estimates. The converse is true for deep neural networks. In this paper, we propose a hybrid model that incorporates the benefits of both approaches. Our new method is data-driven and scalable via a latent, global, deep component. It also handles uncertainty through a local classical Gaussian Process model. Our experiments demonstrate that our method obtains higher accuracy than state-of-the-art methods.
ISSN:2331-8422