New LBS framework unifies LLMs and SSMs for multimodal time-series forecasting

Researchers have developed a new probabilistic framework called LLM-integrated Bayesian State Space Models (LBS) for multimodal time-series forecasting. This framework combines a state space model (SSM) backbone with a pretrained large language model (LLM) to jointly forecast numerical and textual data. LBS allows for flexible input and output horizons, provides uncertainty quantification, and improves temporal generalization. Experiments on the TextTimeCorpus benchmark showed LBS outperforming the previous state-of-the-art by 13.20% and generating human-readable forecast summaries. AI

IMPACT This research offers a new foundation for multimodal temporal reasoning, potentially improving forecasting accuracy and interpretability in domains with both numerical and textual data.

RANK_REASON The cluster describes a new academic paper detailing a novel model architecture for time-series forecasting. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Hugging Face
• Large Language Model
• LLM-Integrated Bayesian State Space Models
• London Business School
• State Space Model
• TextTimeCorpus

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →