Saved in:
Bibliographic Details
Main Authors: Malacarne, Sara, Hoel-Høiseth, Eirik, Aune, Erlend, Biro, David Zsolt, Ruocco, Massimiliano
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2604.27172
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909001608855552
author Malacarne, Sara
Hoel-Høiseth, Eirik
Aune, Erlend
Biro, David Zsolt
Ruocco, Massimiliano
author_facet Malacarne, Sara
Hoel-Høiseth, Eirik
Aune, Erlend
Biro, David Zsolt
Ruocco, Massimiliano
contents We propose C-MTAD-GAT, an \emph{unsupervised}, \emph{context-aware} graph-attention model for anomaly detection in multivariate time series from mobile networks. C-MTAD-GAT combines graph attention with lightweight context embeddings, and uses a deterministic reconstruction head and multi-step forecaster to produce anomaly scores. Detection thresholds are calibrated \emph{without labels} from validation residuals, keeping the pipeline fully unsupervised. On the public TELCO dataset, C-MTAD-GAT consistently outperforms MTAD-GAT and the Telco-specific DC-VAE, two state-of-the-art baselines, in both event-level and pointwise F1, while triggering substantially fewer alarms. C-MTAD-GAT is also deployed in the Core network of a national mobile operator, demonstrating its resilience in real industrial settings.
format Preprint
id arxiv_https___arxiv_org_abs_2604_27172
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Context-Aware Graph Attention for Unsupervised Telco Anomaly Detection
Malacarne, Sara
Hoel-Høiseth, Eirik
Aune, Erlend
Biro, David Zsolt
Ruocco, Massimiliano
Machine Learning
We propose C-MTAD-GAT, an \emph{unsupervised}, \emph{context-aware} graph-attention model for anomaly detection in multivariate time series from mobile networks. C-MTAD-GAT combines graph attention with lightweight context embeddings, and uses a deterministic reconstruction head and multi-step forecaster to produce anomaly scores. Detection thresholds are calibrated \emph{without labels} from validation residuals, keeping the pipeline fully unsupervised. On the public TELCO dataset, C-MTAD-GAT consistently outperforms MTAD-GAT and the Telco-specific DC-VAE, two state-of-the-art baselines, in both event-level and pointwise F1, while triggering substantially fewer alarms. C-MTAD-GAT is also deployed in the Core network of a national mobile operator, demonstrating its resilience in real industrial settings.
title Context-Aware Graph Attention for Unsupervised Telco Anomaly Detection
topic Machine Learning
url https://arxiv.org/abs/2604.27172