Parameter Efficient Multi-Class Intelligent Scheduling for Multimodal Online Distributed Industrial Anomaly Detection
Researchers have introduced a new framework called MODIAD for multimodal online distributed industrial anomaly detection. This framework addresses the limitations of existing methods by focusing on real-world industrial environments with continuously generated, distributed data. MODIAD includes a Multi-class Intelligent Scheduling problem and a Sequential Marginal Gain Greedy algorithm to manage model updates efficiently under resource constraints. Additionally, a Resource Efficient Class-Wise Low Rank Adaptation (REC-LoRA) strategy is proposed to reduce computational and communication overhead while maintaining detection performance. AI
IMPACT Introduces a novel approach to industrial anomaly detection, potentially improving efficiency and performance in real-world distributed systems.