PulseAugur
LIVE 01:42:54
tool · [1 source] ·
0
tool

Researchers develop new method for cross-paradigm graph backdoor attacks using promptable subgraph triggers.

Researchers have developed a new method called Cross-Paradigm Graph Backdoor Attacks with Promptable Subgraph Triggers (CP-GBA) to address vulnerabilities in Graph Neural Networks (GNNs). Existing attacks are often limited to specific GNN learning paradigms, hindering their effectiveness across different frameworks. CP-GBA utilizes Graph Prompt Learning to create transferable subgraph triggers that are class-aware, feature-rich, and structurally sound, demonstrating state-of-the-art attack success rates in experiments. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT This research highlights new attack vectors against GNNs, potentially influencing the development of more robust defenses.

RANK_REASON This is a research paper detailing a new method for backdoor attacks on Graph Neural Networks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

COVERAGE [1]

  1. arXiv cs.LG TIER_1 · Dongyi Liu, Jiangtong Li ·

    Cross-Paradigm Graph Backdoor Attacks with Promptable Subgraph Triggers

    arXiv:2510.22555v2 Announce Type: replace-cross Abstract: Graph Neural Networks(GNNs) are vulnerable to backdoor attacks, where adversaries implant malicious triggers to manipulate model predictions. Existing trigger generators are often simplistic in structure and overly reliant…