TS-RAG: Retrieval-Augmented Generation based Time Series Foundation Models are Stronger Zero-Shot Forecaster

2025-03-06

AI Summary

Existing time-series foundation models (TSFMs) suffer from poor generalization and limited interpretability in zero-shot forecasting. To address this, we propose TS-RAG—the first retrieval-augmented generation framework tailored for zero-shot time-series forecasting. Our method leverages a pre-trained encoder to construct a zero-shot inference pipeline: (1) it introduces the RAG paradigm to time-series forecasting for the first time, enabling semantic retrieval of historically similar temporal patterns; (2) it incorporates a learnable Mixture-of-Experts (MoE) fusion module that dynamically selects relevant patterns and aligns cross-series semantics without fine-tuning. Evaluated on seven public benchmarks, TS-RAG achieves a new zero-shot state-of-the-art, outperforming prior TSFMs by an average of 6.51% in forecasting accuracy. Crucially, it simultaneously provides traceable, human-interpretable retrieval evidence—thereby bridging performance gains with model transparency.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Recently, Large Language Models (LLMs) and Foundation Models (FMs) have become prevalent for time series forecasting tasks. However, fine-tuning large language models (LLMs) for forecasting enables the adaptation to specific domains but may not generalize well across diverse, unseen datasets. Meanwhile, existing time series foundation models (TSFMs) lack inherent mechanisms for domain adaptation and suffer from limited interpretability, making them suboptimal for zero-shot forecasting. To this end, we present TS-RAG, a retrieval-augmented generation based time series forecasting framework that enhances the generalization capability and interpretability of TSFMs. Specifically, TS-RAG leverages pre-trained time series encoders to retrieve semantically relevant time series segments from a dedicated knowledge database, incorporating contextual patterns for the given time series query. Next, we develop a learnable Mixture-of-Experts (MoE)-based augmentation module, which dynamically fuses retrieved time series patterns with the TSFM's representation of the input query, improving forecasting accuracy without requiring task-specific fine-tuning. Thorough empirical studies on seven public benchmark datasets demonstrate that TS-RAG achieves state-of-the-art zero-shot forecasting performance, outperforming TSFMs by up to 6.51% across diverse domains and showcasing desired interpretability.

Problem

• Enhance generalization in time series forecasting.
• Improve interpretability of time series models.
• Achieve zero-shot forecasting without fine-tuning.

Innovation

• Retrieval-augmented generation for time series
• Mixture-of-Experts module for dynamic fusion
• Pre-trained encoders for semantic retrieval