ALARM: Automated MLLM-Based Anomaly Detection in Complex-EnviRonment Monitoring with Uncertainty Quantification

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Congjing, Lin, Feng, Zhao, Xinyi, Guo, Pei, Li, Wei, Chen, Lin, Zhao, Chaoyue, Huang, Shuai
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914363867136000
author Zhang, Congjing
Lin, Feng
Zhao, Xinyi
Guo, Pei
Li, Wei
Chen, Lin
Zhao, Chaoyue
Huang, Shuai
author_facet Zhang, Congjing
Lin, Feng
Zhao, Xinyi
Guo, Pei
Li, Wei
Chen, Lin
Zhao, Chaoyue
Huang, Shuai
contents The advance of Large Language Models (LLMs) has greatly stimulated research interest in developing multi-modal LLM (MLLM)-based visual anomaly detection (VAD) algorithms that can be deployed in complex environments. The challenge is that in these complex environments, the anomalies are sometimes highly contextual and also ambiguous, and thereby, uncertainty quantification (UQ) is a crucial capacity for an MLLM-based VAD system to succeed. In this paper, we introduce our UQ-supported MLLM-based VAD framework called ALARM. ALARM integrates UQ with quality-assurance techniques like reasoning chain, self-reflection, and MLLM ensemble for robust and accurate performance and is designed based on a rigorous probabilistic inference pipeline and computational process. Extensive empirical evaluations are conducted using the real-world smart-home benchmark data and wound image classification data, which shows ALARM's superior performance and its generic applicability across different domains for reliable decision-making.
format Preprint
id arxiv_https___arxiv_org_abs_2512_03101
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ALARM: Automated MLLM-Based Anomaly Detection in Complex-EnviRonment Monitoring with Uncertainty Quantification
Zhang, Congjing
Lin, Feng
Zhao, Xinyi
Guo, Pei
Li, Wei
Chen, Lin
Zhao, Chaoyue
Huang, Shuai
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
The advance of Large Language Models (LLMs) has greatly stimulated research interest in developing multi-modal LLM (MLLM)-based visual anomaly detection (VAD) algorithms that can be deployed in complex environments. The challenge is that in these complex environments, the anomalies are sometimes highly contextual and also ambiguous, and thereby, uncertainty quantification (UQ) is a crucial capacity for an MLLM-based VAD system to succeed. In this paper, we introduce our UQ-supported MLLM-based VAD framework called ALARM. ALARM integrates UQ with quality-assurance techniques like reasoning chain, self-reflection, and MLLM ensemble for robust and accurate performance and is designed based on a rigorous probabilistic inference pipeline and computational process. Extensive empirical evaluations are conducted using the real-world smart-home benchmark data and wound image classification data, which shows ALARM's superior performance and its generic applicability across different domains for reliable decision-making.
title ALARM: Automated MLLM-Based Anomaly Detection in Complex-EnviRonment Monitoring with Uncertainty Quantification
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.03101