Energy-Based Generative Models for Distributed Acoustic Sensing Event Classification in Telecom Networks

Distributed fiber-optic sensing combined with machine learning enables continuous monitoring of telecom infrastructure. We employ generative modeling for event classification, supporting semi­ supervised learning, uncertainty calibration, and noise resilience. Our approach offers a scalable, data-efficient solution for real-world deployment in complex environments.