Saved in:
Bibliographic Details
Main Authors: Ashok, Arjun, Marcotte, Étienne, Zantedeschi, Valentina, Chapados, Nicolas, Drouin, Alexandre
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2310.01327
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914727030947840
author Ashok, Arjun
Marcotte, Étienne
Zantedeschi, Valentina
Chapados, Nicolas
Drouin, Alexandre
author_facet Ashok, Arjun
Marcotte, Étienne
Zantedeschi, Valentina
Chapados, Nicolas
Drouin, Alexandre
contents We introduce a new model for multivariate probabilistic time series prediction, designed to flexibly address a range of tasks including forecasting, interpolation, and their combinations. Building on copula theory, we propose a simplified objective for the recently-introduced transformer-based attentional copulas (TACTiS), wherein the number of distributional parameters now scales linearly with the number of variables instead of factorially. The new objective requires the introduction of a training curriculum, which goes hand-in-hand with necessary changes to the original architecture. We show that the resulting model has significantly better training dynamics and achieves state-of-the-art performance across diverse real-world forecasting tasks, while maintaining the flexibility of prior work, such as seamless handling of unaligned and unevenly-sampled time series. Code is made available at https://github.com/ServiceNow/TACTiS.
format Preprint
id arxiv_https___arxiv_org_abs_2310_01327
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle TACTiS-2: Better, Faster, Simpler Attentional Copulas for Multivariate Time Series
Ashok, Arjun
Marcotte, Étienne
Zantedeschi, Valentina
Chapados, Nicolas
Drouin, Alexandre
Machine Learning
Artificial Intelligence
We introduce a new model for multivariate probabilistic time series prediction, designed to flexibly address a range of tasks including forecasting, interpolation, and their combinations. Building on copula theory, we propose a simplified objective for the recently-introduced transformer-based attentional copulas (TACTiS), wherein the number of distributional parameters now scales linearly with the number of variables instead of factorially. The new objective requires the introduction of a training curriculum, which goes hand-in-hand with necessary changes to the original architecture. We show that the resulting model has significantly better training dynamics and achieves state-of-the-art performance across diverse real-world forecasting tasks, while maintaining the flexibility of prior work, such as seamless handling of unaligned and unevenly-sampled time series. Code is made available at https://github.com/ServiceNow/TACTiS.
title TACTiS-2: Better, Faster, Simpler Attentional Copulas for Multivariate Time Series
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2310.01327