Saved in:
Bibliographic Details
Main Authors: Dutta, Utsav, Pakazad, Sina Khoshfetrat, Ohlsson, Henrik
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.14543
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908372343717888
author Dutta, Utsav
Pakazad, Sina Khoshfetrat
Ohlsson, Henrik
author_facet Dutta, Utsav
Pakazad, Sina Khoshfetrat
Ohlsson, Henrik
contents Traditional time series models are task-specific and often depend on dataset-specific training and extensive feature engineering. While Transformer-based architectures have improved scalability, foundation models, commonplace in text, vision, and audio, remain under-explored for time series and are largely restricted to forecasting. We introduce $\textbf{CHARM}$, a foundation embedding model for multivariate time series that learns shared, transferable, and domain-aware representations. To address the unique difficulties of time series foundation learning, $\textbf{CHARM}$ incorporates architectural innovations that integrate channel-level textual descriptions while remaining invariant to channel order. The model is trained using a Joint Embedding Predictive Architecture (JEPA), with novel augmentation schemes and a loss function designed to improve interpretability and training stability. Our $7$M-parameter model achieves state-of-the-art performance across diverse downstream tasks, setting a new benchmark for time series representation learning.
format Preprint
id arxiv_https___arxiv_org_abs_2505_14543
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Time to Embed: Unlocking Foundation Models for Time Series with Channel Descriptions
Dutta, Utsav
Pakazad, Sina Khoshfetrat
Ohlsson, Henrik
Machine Learning
Traditional time series models are task-specific and often depend on dataset-specific training and extensive feature engineering. While Transformer-based architectures have improved scalability, foundation models, commonplace in text, vision, and audio, remain under-explored for time series and are largely restricted to forecasting. We introduce $\textbf{CHARM}$, a foundation embedding model for multivariate time series that learns shared, transferable, and domain-aware representations. To address the unique difficulties of time series foundation learning, $\textbf{CHARM}$ incorporates architectural innovations that integrate channel-level textual descriptions while remaining invariant to channel order. The model is trained using a Joint Embedding Predictive Architecture (JEPA), with novel augmentation schemes and a loss function designed to improve interpretability and training stability. Our $7$M-parameter model achieves state-of-the-art performance across diverse downstream tasks, setting a new benchmark for time series representation learning.
title Time to Embed: Unlocking Foundation Models for Time Series with Channel Descriptions
topic Machine Learning
url https://arxiv.org/abs/2505.14543