·
DataBubble
  • Home
  • Models
  • News
  • Compare
  • Boards
  • Pricing
  • About
  • Newsletter
  • Methodology
  • Contact
Latest
Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning10h◆Subagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic Tasks10h◆Distribution-Consistent Inference for Dynamic Sparse Mixture-of-Experts10h◆In RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Document Poisoning10h◆FrontierChallenge: Evaluating Scientific Workflow Completion10h◆IBIB: A Protocol for Measuring Enterprise AI Systems by Serving Route, Not Model Identifier10h◆AgenticGen: Reward-Guided Agentic Video Generation for Advertising10h◆Strangers to Themselves: What Language Models Say About Themselves Is Generic10h◆Omni Interaction Agent Technical Report10h◆CoGReV: A Confidence-Gated Post-Hoc Non-Monotonic Belief Revision Framework for Phishing Website Classification10h◆Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability10h◆Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization10h◆RiLM: Parameter-Efficient Language Modeling via Geodesic Decoding10h◆Tracing Computation Density in LLMs10h◆Cultural Binding Heads in Language Models10h◆Direct Diversity Optimization for Diverse Successful Trajectories in Preference Post-Training10h◆Left-Branching Transformers Excel at Right-Branching Languages: Data Shapes Word Order Preferences in Language Models10h◆Palmyra x6 Technical Report: An Agentic, Tool-Use Model Post-Trained via Anchored Supervised Fine-Tuning10h◆'Ghaib in Translation' aka Unseen Harm: Measuring Cross-Script Safety Inconsistency with 'Missed-in-Urdu' Scores in LLM Hate Speech Detection10h◆DexterSQL: Deep Schema Exploration and Rule-based Correction for Text-to-SQL Generation10h◆Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning10h◆Subagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic Tasks10h◆Distribution-Consistent Inference for Dynamic Sparse Mixture-of-Experts10h◆In RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Document Poisoning10h◆FrontierChallenge: Evaluating Scientific Workflow Completion10h◆IBIB: A Protocol for Measuring Enterprise AI Systems by Serving Route, Not Model Identifier10h◆AgenticGen: Reward-Guided Agentic Video Generation for Advertising10h◆Strangers to Themselves: What Language Models Say About Themselves Is Generic10h◆Omni Interaction Agent Technical Report10h◆CoGReV: A Confidence-Gated Post-Hoc Non-Monotonic Belief Revision Framework for Phishing Website Classification10h◆Bit-Flip Attacks on Vision-Language-Action Models: Action-Decoding Architecture Shapes the Vulnerability10h◆Can Foundation Models Moderate Online Content? Evaluating Instruction- vs. Example-Driven Policy Operationalization10h◆RiLM: Parameter-Efficient Language Modeling via Geodesic Decoding10h◆Tracing Computation Density in LLMs10h◆Cultural Binding Heads in Language Models10h◆Direct Diversity Optimization for Diverse Successful Trajectories in Preference Post-Training10h◆Left-Branching Transformers Excel at Right-Branching Languages: Data Shapes Word Order Preferences in Language Models10h◆Palmyra x6 Technical Report: An Agentic, Tool-Use Model Post-Trained via Anchored Supervised Fine-Tuning10h◆'Ghaib in Translation' aka Unseen Harm: Measuring Cross-Script Safety Inconsistency with 'Missed-in-Urdu' Scores in LLM Hate Speech Detection10h◆DexterSQL: Deep Schema Exploration and Rule-based Correction for Text-to-SQL Generation10h◆
News/Trapping Attacker in Dilemma: Examining Internal Correlations and External Influences of Trigger for Defending GNN Backdoors
arxiv
PublishedMay 16, 2026 at 4:00 AM
—neutral

Trapping Attacker in Dilemma: Examining Internal Correlations and External Influences of Trigger for Defending GNN Backdoors

Source
arxiv.orgfull article ↗
Read on arxiv→
Publisher summary· verbatim

arXiv:2605.08278v2 Announce Type: replace-cross Abstract: GNNs have become a standard tool for learning on relational data, yet they remain highly vulnerable to backdoor attacks. Prior defenses often depend on inspecting specific subgraph patterns or node features, and thus can be circumvented by ad

Stay posted· Newsletter

A 5-min weekly brief — top movers, price watch, story of the week.

// no spam · unsubscribe one-click · free forever

Discussion
Source
↗
arxiv
Read original ↗All from arxiv →
Tags
04
#graph-neural-networks#backdoor-attacks#security#machine-learning

No replies yet. Be first.

Source
↗
arxiv
Read original ↗All from arxiv →
Tags
04
#graph-neural-networks#backdoor-attacks#security#machine-learning

Related coverage

More from ARXIV
arxivBringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning10harxivSubagents vs Agent Skills: Executing Reusable Knowledge for Long-Horizon Agentic Tasks10harxivDistribution-Consistent Inference for Dynamic Sparse Mixture-of-Experts10harxivIn RAG We Trust? Measuring Robustness of Retrieval-Augmented Generation Under Document Poisoning10h
The Bubble Brief
WEEKLY

Read graph-neural-networks insights every Tuesday — top movers, new releases, story of the week.

// no spam · unsubscribe one-click · free forever

Originally published on arxiv ↗
HomeModelsNews