Saved in:
Bibliographic Details
Main Authors: Yang, Haonan, Tang, Jianchao, Li, Zhuo
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2601.16632
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912987258814464
author Yang, Haonan
Tang, Jianchao
Li, Zhuo
author_facet Yang, Haonan
Tang, Jianchao
Li, Zhuo
contents Time series forecasting has witnessed significant progress with deep learning. While prevailing approaches enhance forecasting performance by modifying architectures or introducing novel enhancement strategies, they often fail to dynamically disentangle and leverage the complex, intertwined temporal patterns inherent in time series, thus resulting in the learning of static, averaged representations that lack context-aware capabilities. To address this, we propose the Dual-Prototype Adaptive Disentanglement framework (DPAD), a model-agnostic auxiliary method that equips forecasting models with the ability of pattern disentanglement and context-aware adaptation. Specifically, we construct a Dynamic Dual-Prototype bank (DDP), comprising a common pattern bank with strong temporal priors to capture prevailing trend or seasonal patterns, and a rare pattern bank dynamically memorizing critical yet infrequent events, and then an Dual-Path Context-aware routing (DPC) mechanism is proposed to enhance outputs with selectively retrieved context-specific pattern representations from the DDP. Additionally, we introduce a Disentanglement-Guided Loss (DGLoss) to ensure that each prototype bank specializes in its designated role while maintaining comprehensive coverage. Comprehensive experiments demonstrate that DPAD consistently improves forecasting performance and reliability of state-of-the-art models across diverse real-world benchmarks.
format Preprint
id arxiv_https___arxiv_org_abs_2601_16632
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Dual-Prototype Disentanglement: A Context-Aware Enhancement Framework for Time Series Forecasting
Yang, Haonan
Tang, Jianchao
Li, Zhuo
Machine Learning
Artificial Intelligence
Time series forecasting has witnessed significant progress with deep learning. While prevailing approaches enhance forecasting performance by modifying architectures or introducing novel enhancement strategies, they often fail to dynamically disentangle and leverage the complex, intertwined temporal patterns inherent in time series, thus resulting in the learning of static, averaged representations that lack context-aware capabilities. To address this, we propose the Dual-Prototype Adaptive Disentanglement framework (DPAD), a model-agnostic auxiliary method that equips forecasting models with the ability of pattern disentanglement and context-aware adaptation. Specifically, we construct a Dynamic Dual-Prototype bank (DDP), comprising a common pattern bank with strong temporal priors to capture prevailing trend or seasonal patterns, and a rare pattern bank dynamically memorizing critical yet infrequent events, and then an Dual-Path Context-aware routing (DPC) mechanism is proposed to enhance outputs with selectively retrieved context-specific pattern representations from the DDP. Additionally, we introduce a Disentanglement-Guided Loss (DGLoss) to ensure that each prototype bank specializes in its designated role while maintaining comprehensive coverage. Comprehensive experiments demonstrate that DPAD consistently improves forecasting performance and reliability of state-of-the-art models across diverse real-world benchmarks.
title Dual-Prototype Disentanglement: A Context-Aware Enhancement Framework for Time Series Forecasting
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2601.16632