Home > Poster

Poster

Semi-Supervised Anomaly Detection With Imbalanced Data for Failure Detection in Optical Networks (Th1A.24)

Presenter: Songlin Liu, Beijing University of Posts and Telecomm

We proposed an autoencoder-based anomaly-detection for optical failure detection with imbalanced data (<3%), which identifies implicit failure and achieves detection accuracy of 96.8% and F1 value of 0.9224.

Authors:Songlin Liu, Beijing University of Posts and Telecomm / Danshi Wang, Beijing University of Posts and Telecomm / Chunyu Zhang, Beijing University of Posts and Telecomm / Lingling Wang, Beijing University of Posts and Telecomm / Min Zhang, Beijing University of Posts and Telecomm


Poster Presentation

Authentication Required

Access to poster content is available to logged in registered attendees, please login now to participate.