arXiv：cs.AI· Sky Chenwei Wan, Yifei Y. Wang, Tianjun Hou, Xiqing Chang, Aymeric Jan·· 20 小时前
基于神经符号 VLM 智能体的可解释多变量时间序列事件检测框架 SELA
Grammar of the Wave: Towards Explainable Multivariate Time Series Event Detection via Neuro-Symbolic VLM Agents
AI 导读
针对语言引导的时间序列事件检测（TSED），研究者提出知识表示框架 Event Logic Tree（ELT），将自然语言事件描述转换为信号原语上的结构化时序逻辑，并基于此构建神经符号 VLM 智能体框架 SELA，迭代地从信号可视化中 grounding 原语并按 ELT 约束组合，同时输出事件区间与树状解释。
正文
Abstract:Time Series Event Detection (TSED) aims to localize semantically meaningful events in time series data, with critical applications in high-stakes domains. Unlike statistical anomalies, events are often defined by natural-language descriptions with internal temporal-logic structures across multiple physical channels. However, in real-world settings, dense event annotations are expensive to obtain, making purely supervised learning difficult. We introduce Language-guided TSED, a setting where a model is given textual event descriptions and must ground them to intervals in multivariate signals with little or no labeled data. To address this problem, we propose Event Logic Tree (ELT), a knowledge representation framework that converts linguistic descriptions into structured temporal logic over signal primitives. Building on ELT, we present SELA, a neuro-symbolic VLM agent framework that iteratively grounds primitives from signal visualizations and composes them under ELT constraints, producing both event intervals and faithful tree-structured explanations. We further release a real-world benchmark across energy and climate domains with expert knowledge and annotations. Experiments show that SELA improves over supervised fine-tuning and existing zero/few-shot time series reasoning baselines.
| Comments: | Long paper accepted to EMNLP 2026 main conference, camera-ready version |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA) |
| Cite as: | arXiv:2603.11479 [cs.LG] |
| (or arXiv:2603.11479v5 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2603.11479 arXiv-issued DOI via DataCite |
Submission history
From: Sky Chenwei Wan [view email]
[v1]
Thu, 12 Mar 2026 03:01:48 UTC (242 KB)
[v2]
Tue, 26 May 2026 13:03:04 UTC (1,186 KB)
[v3]
Wed, 10 Jun 2026 22:09:21 UTC (1,187 KB)
[v4]
Sun, 30 Aug 2026 01:09:56 UTC (1,192 KB)
[v5]
Mon, 12 Oct 2026 19:58:13 UTC (1,192 KB)
来源：arXiv：cs.AI · arxiv.org