AltTS: A Dual-Path Framework with Alternating Optimization for Multivariate Time Series Forecasting

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Yuan, Zhihang, Liu, Zhiyuan, Marina, Mahesh K.
Formato: Preprint
Publicado: 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866914324280246272
author Yuan, Zhihang
Liu, Zhiyuan
Marina, Mahesh K.
author_facet Yuan, Zhihang
Liu, Zhiyuan
Marina, Mahesh K.
contents Multivariate time series forecasting involves two qualitatively distinct factors: (i) stable within-series autoregressive (AR) dynamics, and (ii) intermittent cross-dimension interactions that can become spurious over long horizons. We argue that fitting a single model to capture both effects creates an optimization conflict: the high-variance updates needed for cross-dimension modeling can corrupt the gradients that support autoregression, resulting in brittle training and degraded long-horizon accuracy. To address this, we propose ALTTS, a dual-path framework that explicitly decouples autoregression and cross-relation (CR) modeling. In ALTTS, the AR path is instantiated with a linear predictor, while the CR path uses a Transformer equipped with Cross-Relation Self-Attention (CRSA); the two branches are coordinated via alternating optimization to isolate gradient noise and reduce cross-block interference. Extensive experiments on multiple benchmarks show that ALTTS consistently outperforms prior methods, with the most pronounced improvements on long-horizon forecasting. Overall, our results suggest that carefully designed optimization strategies, rather than ever more complex architectures, can be a key driver of progress in multivariate time series forecasting.
format Preprint
id arxiv_https___arxiv_org_abs_2602_11533
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle AltTS: A Dual-Path Framework with Alternating Optimization for Multivariate Time Series Forecasting
Yuan, Zhihang
Liu, Zhiyuan
Marina, Mahesh K.
Machine Learning
Artificial Intelligence
Multivariate time series forecasting involves two qualitatively distinct factors: (i) stable within-series autoregressive (AR) dynamics, and (ii) intermittent cross-dimension interactions that can become spurious over long horizons. We argue that fitting a single model to capture both effects creates an optimization conflict: the high-variance updates needed for cross-dimension modeling can corrupt the gradients that support autoregression, resulting in brittle training and degraded long-horizon accuracy. To address this, we propose ALTTS, a dual-path framework that explicitly decouples autoregression and cross-relation (CR) modeling. In ALTTS, the AR path is instantiated with a linear predictor, while the CR path uses a Transformer equipped with Cross-Relation Self-Attention (CRSA); the two branches are coordinated via alternating optimization to isolate gradient noise and reduce cross-block interference. Extensive experiments on multiple benchmarks show that ALTTS consistently outperforms prior methods, with the most pronounced improvements on long-horizon forecasting. Overall, our results suggest that carefully designed optimization strategies, rather than ever more complex architectures, can be a key driver of progress in multivariate time series forecasting.
title AltTS: A Dual-Path Framework with Alternating Optimization for Multivariate Time Series Forecasting
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2602.11533