Training-Free Time Series Classification via In-Context Reasoning with LLM Agents

2025-10-07

AI Summary

To address the scarcity of labeled data and the suboptimal performance of zero-shot methods in time series classification (TSC), this paper proposes FETA—a training-free, multi-agent framework. FETA introduces a novel three-stage collaborative mechanism: channel-wise decomposition, exemplar retrieval, and large language model (LLM)-based reasoning. It retrieves intra-class prototypical samples via structural similarity matching at the channel level, then leverages LLM contextual reasoning with confidence-weighted fusion to achieve high-accuracy zero-shot classification—while preserving controllable input length. Fully parameter-free, FETA offers strong interpretability and plug-and-play deployment. Evaluated on nine UEA benchmark datasets, FETA significantly outperforms state-of-the-art trainable baselines, establishing new state-of-the-art performance for zero-shot TSC.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Multiagent Systems: Multiagent Learning
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling

Abstract

Time series classification (TSC) spans diverse application scenarios, yet labeled data are often scarce, making task-specific training costly and inflexible. Recent reasoning-oriented large language models (LLMs) show promise in understanding temporal patterns, but purely zero-shot usage remains suboptimal. We propose FETA, a multi-agent framework for training-free TSC via exemplar-based in-context reasoning. FETA decomposes a multivariate series into channel-wise subproblems, retrieves a few structurally similar labeled examples for each channel, and leverages a reasoning LLM to compare the query against these exemplars, producing channel-level labels with self-assessed confidences; a confidence-weighted aggregator then fuses all channel decisions. This design eliminates the need for pretraining or fine-tuning, improves efficiency by pruning irrelevant channels and controlling input length, and enhances interpretability through exemplar grounding and confidence estimation. On nine challenging UEA datasets, FETA achieves strong accuracy under a fully training-free setting, surpassing multiple trained baselines. These results demonstrate that a multi-agent in-context reasoning framework can transform LLMs into competitive, plug-and-play TSC solvers without any parameter training. The code is available at https://github.com/SongyuanSui/FETATSC.

Problem

• Classifying time series without training using in-context reasoning
• Addressing scarce labeled data via multi-agent exemplar comparison
• Eliminating pretraining needs while maintaining competitive accuracy

Innovation

• Multi-agent framework enables training-free time series classification
• Decomposes multivariate series into channel-wise subproblems for analysis
• Uses confidence-weighted aggregation to fuse channel-level decisions