Saved in:
Bibliographic Details
Main Authors: Wu, Yuemin, Wu, Zhongze, Su, Xiu, Yang, Feng, Xu, Hongyan, Lin, Xi, Huang, Wenti, You, Shan, Xu, Chang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2509.14642
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909794986622976
author Wu, Yuemin
Wu, Zhongze
Su, Xiu
Yang, Feng
Xu, Hongyan
Lin, Xi
Huang, Wenti
You, Shan
Xu, Chang
author_facet Wu, Yuemin
Wu, Zhongze
Su, Xiu
Yang, Feng
Xu, Hongyan
Lin, Xi
Huang, Wenti
You, Shan
Xu, Chang
contents Modeling dynamic temporal dependencies is a critical challenge in time series pre-training, which evolve due to distribution shifts and multi-scale patterns. This temporal variability severely impairs the generalization of pre-trained models to downstream tasks. Existing frameworks fail to capture the complex interactions of short- and long-term dependencies, making them susceptible to spurious correlations that degrade generalization. To address these limitations, we propose DeCoP, a Dependency Controlled Pre-training framework that explicitly models dynamic, multi-scale dependencies by simulating evolving inter-patch dependencies. At the input level, DeCoP introduces Instance-wise Patch Normalization (IPN) to mitigate distributional shifts while preserving the unique characteristics of each patch, creating a robust foundation for representation learning. At the latent level, a hierarchical Dependency Controlled Learning (DCL) strategy explicitly models inter-patch dependencies across multiple temporal scales, with an Instance-level Contrastive Module (ICM) enhances global generalization by learning instance-discriminative representations from time-invariant positive pairs. DeCoP achieves state-of-the-art results on ten datasets with lower computing resources, improving MSE by 3% on ETTh1 over PatchTST using only 37% of the FLOPs.
format Preprint
id arxiv_https___arxiv_org_abs_2509_14642
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle DeCoP: Enhancing Self-Supervised Time Series Representation with Dependency Controlled Pre-training
Wu, Yuemin
Wu, Zhongze
Su, Xiu
Yang, Feng
Xu, Hongyan
Lin, Xi
Huang, Wenti
You, Shan
Xu, Chang
Machine Learning
Artificial Intelligence
Modeling dynamic temporal dependencies is a critical challenge in time series pre-training, which evolve due to distribution shifts and multi-scale patterns. This temporal variability severely impairs the generalization of pre-trained models to downstream tasks. Existing frameworks fail to capture the complex interactions of short- and long-term dependencies, making them susceptible to spurious correlations that degrade generalization. To address these limitations, we propose DeCoP, a Dependency Controlled Pre-training framework that explicitly models dynamic, multi-scale dependencies by simulating evolving inter-patch dependencies. At the input level, DeCoP introduces Instance-wise Patch Normalization (IPN) to mitigate distributional shifts while preserving the unique characteristics of each patch, creating a robust foundation for representation learning. At the latent level, a hierarchical Dependency Controlled Learning (DCL) strategy explicitly models inter-patch dependencies across multiple temporal scales, with an Instance-level Contrastive Module (ICM) enhances global generalization by learning instance-discriminative representations from time-invariant positive pairs. DeCoP achieves state-of-the-art results on ten datasets with lower computing resources, improving MSE by 3% on ETTh1 over PatchTST using only 37% of the FLOPs.
title DeCoP: Enhancing Self-Supervised Time Series Representation with Dependency Controlled Pre-training
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2509.14642