TACTiS-2: Better, Faster, Simpler Attentional Copulas for Multivariate Time Series

TL;DR

This paper presents TACTiS-2, an improved attentional copula model for multivariate time series that is faster, simpler, and more flexible, achieving state-of-the-art results in forecasting tasks.

Contribution

Introduces a simplified, scalable transformer-based attentional copula model with a new training objective and curriculum, enhancing training dynamics and performance.

Findings

Achieves state-of-the-art forecasting performance.

Significantly improved training dynamics.

Handles unaligned and unevenly-sampled time series seamlessly.

Abstract

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.