Saved in:
Bibliographic Details
Main Authors: Tao, Wei, Qu, Xiaoyang, Lu, Kai, Wan, Jiguang, Li, Guokuan, Wang, Jianzong
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2504.09504
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909578626596864
author Tao, Wei
Qu, Xiaoyang
Lu, Kai
Wan, Jiguang
Li, Guokuan
Wang, Jianzong
author_facet Tao, Wei
Qu, Xiaoyang
Lu, Kai
Wan, Jiguang
Li, Guokuan
Wang, Jianzong
contents When applying pre-trained large language models (LLMs) to address anomaly detection tasks, the multivariate time series (MTS) modality of anomaly detection does not align with the text modality of LLMs. Existing methods simply transform the MTS data into multiple univariate time series sequences, which can cause many problems. This paper introduces MADLLM, a novel multivariate anomaly detection method via pre-trained LLMs. We design a new triple encoding technique to align the MTS modality with the text modality of LLMs. Specifically, this technique integrates the traditional patch embedding method with two novel embedding approaches: Skip Embedding, which alters the order of patch processing in traditional methods to help LLMs retain knowledge of previous features, and Feature Embedding, which leverages contrastive learning to allow the model to better understand the correlations between different features. Experimental results demonstrate that our method outperforms state-of-the-art methods in various public anomaly detection datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2504_09504
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle MADLLM: Multivariate Anomaly Detection via Pre-trained LLMs
Tao, Wei
Qu, Xiaoyang
Lu, Kai
Wan, Jiguang
Li, Guokuan
Wang, Jianzong
Computation and Language
When applying pre-trained large language models (LLMs) to address anomaly detection tasks, the multivariate time series (MTS) modality of anomaly detection does not align with the text modality of LLMs. Existing methods simply transform the MTS data into multiple univariate time series sequences, which can cause many problems. This paper introduces MADLLM, a novel multivariate anomaly detection method via pre-trained LLMs. We design a new triple encoding technique to align the MTS modality with the text modality of LLMs. Specifically, this technique integrates the traditional patch embedding method with two novel embedding approaches: Skip Embedding, which alters the order of patch processing in traditional methods to help LLMs retain knowledge of previous features, and Feature Embedding, which leverages contrastive learning to allow the model to better understand the correlations between different features. Experimental results demonstrate that our method outperforms state-of-the-art methods in various public anomaly detection datasets.
title MADLLM: Multivariate Anomaly Detection via Pre-trained LLMs
topic Computation and Language
url https://arxiv.org/abs/2504.09504