PulseAugur
EN
LIVE 08:10:33

New adversarial attack targets GNN-based anomaly detection in sensor networks

By PulseAugur Editorial · [1 sources] ·

Researchers have developed BETA, a novel indirect adversarial attack designed to compromise graph neural network (GNN) based anomaly detection systems in sensor networks. This attack method allows an adversary to perturb readings from a limited set of nodes, excluding the target sensor, to manipulate anomaly detection outcomes. BETA utilizes a graph explanatory model and a centrality-based pruning strategy to identify influential nodes for perturbation, demonstrating significant effectiveness in reducing the F1-score of existing GNN detectors by an average of 36.07% to 50.45% across real-world datasets. AI

IMPACT This research highlights vulnerabilities in GNN-based anomaly detection, potentially leading to more robust security measures for sensor networks.

RANK_REASON The cluster contains an academic paper detailing a new method for attacking graph neural networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

New adversarial attack targets GNN-based anomaly detection in sensor networks

COVERAGE [1]

  1. arXiv cs.LG TIER_1 English(EN) · Sanju Xaviar, Omid Ardakanian ·

    Budgeted Indirect Adversarial Attack on Graph-Based Anomaly Detection in Sensor Networks

    arXiv:2509.17987v2 Announce Type: replace Abstract: Graph Neural Networks (GNNs) have emerged as powerful models for anomaly detection in sensor networks, particularly when analyzing multivariate time series. In this work, we introduce BETA, a novel indirect evasion attack target…

RELATED ENTITIES

RELATED TOPICS