Research Topics | Inspiring Group

Secure and Reliable Infrastructure for LLM-powered AI Agents

Mon, 01 Jan 0001 00:00:00 +0000

Our research systematically enhances AI agents with adversarial robustness, operational reliability, data privacy, and verifiability. By embedding deterministic guarantees into inherently unstable AI agent systems, our infrastructure empowers the large-scale industrial adoption of AI agent applications.

Representative References

BlockA2A: Towards Secure and Verifiable Agent-to-Agent Interoperability , 2025.
Agentic Privacy-Preserving Machine Learning , 2025.
Anonymization-Enhanced Privacy Protection for Mobile GUI Agents: Available but Invisible , 2026.
Blind Gods and Broken Screens: Architecting a Secure, Intent-Centric Mobile Agent Operating System , 2026.

Network Security

Mon, 01 Jan 0001 00:00:00 +0000

Our research focuses on making large-scale networked systems resilient, measurable, smart, and secure by design.

Representative References

Learning with Semantics: Towards a Semantics-Aware Routing Anomaly Detection System . Sec24 , 2024.
Dynamic Network Security Function Enforcement via Joint Flow and Function Scheduling . IEEE Transactions on Information Forensics and Security (TIFS) 2022 , 2022.
CertTA: Certified Robustness Made Practical for Learning-Based Traffic Analysis . Sec25 , 2025.
Brain-on-Switch: Towards Advanced Intelligent Network Dataplane via NN-Driven Traffic Analysis at Line-Speed . NSDI24 , 2024.

Privacy-Preserving Machine Learning

Mon, 01 Jan 0001 00:00:00 +0000

We study algorithmic and systems techniques for private AI pipelines in both centralized and federated settings.

Representative References

Pencil: Private and Extensible Collaborative Learning without Non-Colluding Assumption . NDSS24 , 2024.
CoGNN: Towards Secure and Efficient Collaborative Graph Learning . CCS24 , 2024.
RingSG: Optimal Secure Vertex-Centric Computation for Collaborative Graph Processing . CCS25 , 2025.
Defending Against Data Reconstruction Attacks in Federated Learning: An Information Theory Approach . Sec24 , 2024.

Blockchain Infrastructures and Web3

Mon, 01 Jan 0001 00:00:00 +0000

Our work targets both theoretical guarantees and practical deployment for decentralized infrastructures.

Representative References

HyperService: Interoperability and Programmability Across Heterogeneous Blockchains . CCS19 , 2019.
Make Web3.0 Connected . IEEE Transactions on Dependable and Secure Computing (TDSC) 2022 , 2022.
martFL: Enabling Utility-Driven Data Marketplace with a Robust and Verifiable Federated Learning Architecture . CCS23 , 2023.