Pith. sign in

REVIEW 4 cited by

Graph Neural Network-Based Anomaly Detection in Multivariate Time Series

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2106.06947 v1 pith:7OVQGL55 submitted 2021-06-13 cs.LG cs.AI

Graph Neural Network-Based Anomaly Detection in Multivariate Time Series

classification cs.LG cs.AI
keywords anomaliesanomalyrelationshipsseriestimeaccuratelyapproachapproaches
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Given high-dimensional time series data (e.g., sensor data), how can we detect anomalous events, such as system faults and attacks? More challengingly, how can we do this in a way that captures complex inter-sensor relationships, and detects and explains anomalies which deviate from these relationships? Recently, deep learning approaches have enabled improvements in anomaly detection in high-dimensional datasets; however, existing methods do not explicitly learn the structure of existing relationships between variables, or use them to predict the expected behavior of time series. Our approach combines a structure learning approach with graph neural networks, additionally using attention weights to provide explainability for the detected anomalies. Experiments on two real-world sensor datasets with ground truth anomalies show that our method detects anomalies more accurately than baseline approaches, accurately captures correlations between sensors, and allows users to deduce the root cause of a detected anomaly.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 4 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. ECoLAD: Selecting Anomaly Detectors for Automotive Deployment via Compute-Reduction Evaluation

    cs.LG 2026-03 conditional novelty 7.0

    ECoLAD shows classical anomaly detectors maintain coverage and accuracy lift under automotive compute limits while several deep methods lose feasibility first.

  2. Modeling Normal Is All You Need: Joint Latent Clustering for Anomaly Detection in Multimodal Cyber-Physical Systems

    cs.LG 2026-07 conditional novelty 6.0

    A VaDE-based latent clustering detector that drops reconstruction wins a fair, difficulty-stratified anomaly detection protocol on three CPS datasets, with margins tracking dataset multimodality.

  3. Surprised by Attention: Predictable Query Dynamics for Time Series Anomaly Detection

    cs.LG 2026-03 conditional novelty 6.0

    Predicting multi-head attention queries from history and scoring cosine mismatch against an EMA target, combined with reconstruction error, improves unsupervised multivariate anomaly ranking and localization.

  4. ECoLAD: Selecting Anomaly Detectors for Automotive Deployment via Compute-Reduction Evaluation

    cs.LG 2026-03 conditional novelty 6.0

    Under a monotone compute-reduction ladder with thread caps, classical anomaly detectors remain throughput-feasible for automotive telemetry while several deep methods lose coverage first.