QuiZSF: A Retrieval-Augmented Framework for Zero-Shot Time Series Forecasting

TL;DR

QuiZSF presents a retrieval-augmented framework for zero-shot time-series forecasting that unifies scalable temporal retrieval with forecasting through three core modules: ChronoRAG Base (CRB) for efficient cross-domain retrieval, Multi-grained Series Interaction Learner (MSIL) for rich target-retrieved sequence interactions, and Model Cooperation Coherer (MCC) for modality-aware integration with both Non-LLM and LLM-based TSPMs. The framework employs a hierarchical, prototype-based ChronoRAG Base with Hybrid and Hierarchical Time-series Retrieval (HHTR), enabling domain-aware, scalable retrieval. Through comprehensive experiments on five public benchmarks, QuiZSF achieves state-of-the-art zero-shot performance, ranking first in up to 87.5% of settings while maintaining efficiency, and demonstrates strong generalization to unseen domains. The work extends retrieval-augmented AI to time series, offering a scalable approach for adaptive forecasting in dynamic Web environments.

Abstract

Accurate forecasting of sequential data streams is a cornerstone of modern Web services, supporting applications such as traffic management, user behavior modeling, and online anomaly prevention. However, in many Web environments, new domains emerge rapidly and labeled history data is scarce, which makes zero-shot forecasting particularly challenging. Existing time-series pre-trained models (TSPMs) show promise but they lack the ability to dynamically incorporate external knowledge, while conventional retrieval-augmented generation (RAG) methods are rarely extended beyond text. In this work, we present \textbf{QuiZSF}, a retrieval-augmented forecasting framework that integrates search and forecasting for time series data. The framework performs search by retrieving structurally similar sequences from a large-scale time-series database, and it performs forecasting by integrating the retrieved knowledge into the target sequence. Specifically, QuiZSF introduces a \textbf{ChronoRAG Base}, a hierarchical tree-structured database that enables scalable and domain-aware retrieval, a \textbf{Multi-grained Series Interaction Learner} that captures fine- and coarse-grained dependencies between target and retrieved sequences, and a \textbf{Model Cooperation Coherer} that adapts retrieved knowledge to TSPMs. This design teaches models to actively perform search, align auxiliary information across modalities, and leverage it for more accurate forecasting. Extensive experiments on five public benchmarks demonstrate that QuiZSF consistently outperforms strong baselines, ranking first in up to \textbf{87.5\%} of zero-shot forecasting settings while maintaining high efficiency.

Figures (10)

• Figure 1: Motivation. Time series across domains often exhibit similar temporal patterns, which can be retrieved and reused as auxiliary knowledge.
• Figure 2: Overview of QuiZSF.
• Figure 3: (a) HHTR: Domain-aware and global retrieval via a hierarchical index. (b) MSIL: Interaction and average patterns extracted from retrieved sequences are fused with the target via cross-attention.
• Figure 4: (a) Size and time overview of QuiZTST vs. pre-trained TS benchmarks. Plot each model based on model size and the CPU inference time per batch. (b) Hyperparameter analysis on ETTm1 -> ETTm2.
• Figure 5: Case studies on ETTh2 prediction.