aiqbench analysis
TiMi advances multimodal time series forecasting by leveraging Large Language Models to generate causal inferences from textual events, effectively guiding numerical predictions without requiring explicit representation alignment. The proposed Multimodal Mixture-of-Experts module serves as a lightweight plug-in that consistently outperforms existing baselines across sixteen real-world benchmarks while enhancing model interpretability. This approach demonstrates how integrating LLM reasoning capabilities can significantly improve the accuracy of complex forecasting tasks involving exogenous factors like policy announcements or emergency reports.
TiMi: Empower Time Series Transformers with Multimodal Mixture of Experts
arXiv:2602.21693v2 Announce Type: replace Abstract: Multimodal time series forecasting has garnered significant attention for its potential to provide more accurate predictions than traditional single-modality models by leveraging rich information inherent in other modalities. However, due to fundamental challenges in modality alignment, existing methods often struggle to effectively incorporate multimodal data into predictions, particularly textual information that has a causal influence on time series fluctuations, such as emergency reports and policy announcements. In this paper, we reflect on the role of textual information in numerical forecasting and propose Time series transformers with Multimodal Mixture-of-Experts, TiMi, to unleash the causal reasoning capabilities of LLMs. Concretely, TiMi utilizes LLMs to generate inferences on future developments, which serve as guidance for time series forecasting. To seamlessly integrate both exogenous factors and time series into predictions, we introduce a Multimodal Mixture-of-Experts (MMoE) module as a lightweight plug-in to empower Transformer-based time series models for multimodal forecasting, eliminating the need for explicit representation-level alignment. Experimentally, our proposed TiMi demonstrates consistent state-of-the-art performance on sixteen real-world multimodal forecasting benchmarks, outperforming advanced baselines while offering both strong adaptability and interpretability.
Abstract:Multimodal time series forecasting has garnered significant attention for its potential to provide more accurate predictions than traditional single-modality models by leveraging rich information inherent in other modalities. However, due to fundamental challenges in modality alignment, existing methods often struggle to effectively incorporate multimodal data into predictions, particularly textual information that has a causal influence on time series fluctuations, such as emergency reports and policy announcements. In this paper, we reflect on the role of textual information in numerical forecasting and propose Time series transformers with Multimodal Mixture-of-Experts, TiMi, to unleash the causal reasoning capabilities of LLMs. Concretely, TiMi utilizes LLMs to generate inferences on future developments, which serve as guidance for time series forecasting. To seamlessly integrate both exogenous factors and time series into predictions, we introduce a Multimodal Mixture-of-Experts (MMoE) module as a lightweight plug-in to empower Transformer-based time series models for multimodal forecasting, eliminating the need for explicit representation-level alignment. Experimentally, our proposed TiMi demonstrates consistent state-of-the-art performance on sixteen real-world multimodal forecasting benchmarks, outperforming advanced baselines while offering both strong adaptability and interpretability.
Machine Learning (cs.LG)
arXiv:2602.21693 [cs.LG]
(or arXiv:2602.21693v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2602.21693 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 269 words
Related coverage
Introducing Command A+: Making sovereign agentic capabilities available to all
Cohere · May 20, 2026
Welcome Muse Spark Open: open-weight frontier multimodal mixture-of-experts model from Meta
Hugging Face · Oct 15, 2026
Vertical Advantage: Transforming Industries with Lakebase and Agentic AI
Databricks · Aug 27, 2026
MedTVL: Harnessing Vision and Language for Medical Time Series Classification
arXiv · Sep 1, 2026