Mixing It Up: Exploring Mixer Networks for Irregular Multivariate Time Series Forecasting

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Klötergens, Christian, Dernedde, Tim, Schmidt-Thieme, Lars, Yalavarthi, Vijaya Krishna
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918356924235776
author Klötergens, Christian
Dernedde, Tim
Schmidt-Thieme, Lars
Yalavarthi, Vijaya Krishna
author_facet Klötergens, Christian
Dernedde, Tim
Schmidt-Thieme, Lars
Yalavarthi, Vijaya Krishna
contents Forecasting irregularly sampled multivariate time series with missing values (IMTS) is a fundamental challenge in domains such as healthcare, climate science, and biology. While recent advances in vision and time series forecasting have shown that lightweight MLP-based architectures (e.g., MLP-Mixer, TSMixer) can rival attention-based models in both accuracy and efficiency, their applicability to irregular and sparse time series remains unexplored. In this paper, we propose IMTS-Mixer, a novel architecture that adapts the principles of Mixer models to the IMTS setting. IMTS-Mixer introduces two key components: (1) ISCAM, a channel-wise encoder that transforms irregular observations into fixed-size vectors using simple MLPs, and (2) ConTP, a continuous time decoder that supports forecasting at arbitrary time points. In our experiments on established benchmark datasets we show that our model achieves state-of-the- art performance in both forecasting accuracy and inference time, while using fewer parameters compared to baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2502_11816
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Mixing It Up: Exploring Mixer Networks for Irregular Multivariate Time Series Forecasting
Klötergens, Christian
Dernedde, Tim
Schmidt-Thieme, Lars
Yalavarthi, Vijaya Krishna
Machine Learning
I.5
Forecasting irregularly sampled multivariate time series with missing values (IMTS) is a fundamental challenge in domains such as healthcare, climate science, and biology. While recent advances in vision and time series forecasting have shown that lightweight MLP-based architectures (e.g., MLP-Mixer, TSMixer) can rival attention-based models in both accuracy and efficiency, their applicability to irregular and sparse time series remains unexplored. In this paper, we propose IMTS-Mixer, a novel architecture that adapts the principles of Mixer models to the IMTS setting. IMTS-Mixer introduces two key components: (1) ISCAM, a channel-wise encoder that transforms irregular observations into fixed-size vectors using simple MLPs, and (2) ConTP, a continuous time decoder that supports forecasting at arbitrary time points. In our experiments on established benchmark datasets we show that our model achieves state-of-the- art performance in both forecasting accuracy and inference time, while using fewer parameters compared to baselines.
title Mixing It Up: Exploring Mixer Networks for Irregular Multivariate Time Series Forecasting
topic Machine Learning
I.5
url https://arxiv.org/abs/2502.11816