Associate professor in cybersecurity

Lab

CRYPTAI LAB Brief Intro

We are keen to deliver the world-leading cybersecurity and AI R&D. The CRYPTAI lab (applied cryptography & AI) was found in 2021, and its vision is to pave the way to applied cryptography, information security and machine learning tools to achieve real-world intelligent security and privacy. The lab, led by Dr. Liang, focuses on the R&D on: (1) data protection mechanisms, in particular practical data protection and sharing for encrypted outsourced data solutions, searchable encryption, updatable encryption, lattice-based encryption/signature (post-quantum cryptography); (2) privacy-preserving machine/federated learning: attacks and countermeasures; and (3) application-driven blockchain (e.g., smart contract, oracles) development.

CRYPTAI’s projects

ORQESTRA (Orchestrating the Operational Deployment of Quantum Resistant Services for Next-Generation Secure Defence Systems and Communications) aims to provide practical, formally verified and efficient implementations of Post-Quantum Crypto (PQC) algorithms and abstractions addressing the current need towards bridging the gap between the foundational primitives and hard mathematical problems of PQC to the operational deployment of cryptosystems that can provide certifiable quantum-level security. The added value of the ORQESTRA’s offerings will be validated in 8 real-world defence-oriented use cases and 2 Proof-of-Concept evaluation testbeds, inducing Free-Space QKD/PQC, with high security, trust, timing and performance requirements, focusing on protecting Land, Maritime and Aerial critical infrastructures and communications.

CRYPTAI’s Publications

Recent and selected papers

  • LADDER: Multi-Objective Backdoor Attack via Evolutionary Algorithm, NDSS (A* conference) 2025.
  • MUDGUARD: Taming Malicious Majorities in Federated Learning using Privacy-preserving Byzantine-robust Clustering, SIGMETRICS (A* conference) 2025.
  • Athena: Accelerating KeySwitch and Bootstrapping for Fully Homomorphic Encryption on CUDA GPU ESORICS (A conference) 2025.
  • Low-Frequency Black-Box Backdoor Attack via Evolutionary Algorithm, WACV (A conference) 2025.
  • d-DSE: Distinct Dynamic Searchable Encryption Resisting Volume Leakage in Encrypted Databases, accepted by USENIX Security (A* conference) 2024.
  • Similar Data is Powerful: Enhancing Inference Attacks on SSE with Volume Leakages, accepted by ESORICS (A conference) 2024.
  • Query Recovery from Easy to Hard: Jigsaw Attack against SSE, accepted by USENIX Security (A* conference) 2024.
  • Inject Less, Recover More: Unlocking the Potential of Document Recovery in Injection Attacks Against SSE, accepted by CSF (A conference) 2024.
  • PrivDNN: A Secure Multi-Party Computation Framework for Deep Learning using Partial DNN Encryption, accepted by PETS (A conference) 2024.
  • High Recovery with Fewer Injections: Practical Binary Volumetric Injection Attacks against Dynamic Searchable Encryption, USENIX Security (A* conference) 2023.
  • CCA-1 Secure Updatable Encryption with Adaptive Security, Asiacrypt (A conference) 2023.
  • The Power of Bamboo: On the Post-Compromise Security for Searchable Symmetric Encryption, NDSS (A* conference) 2023.
  • FLVoogd: Robust And Privacy Preserving Federated Learning, ACML 2022.
  • More is Better (Mostly): On the Backdoor Attacks in Federated Graph Neural Networks, ACSAC (A conference) 2022.
  • MVOC: A Lighter Multi-Client Verifiable Outsourced Computation with Hybrid Homomorphic Encryption, ESORICS (A conference) 2022.
  • Explore More, Reveal More - VAL: Volume and Access Pattern Leakage-abuse Attack with Leaked Documents, ESORICS (A conference) 2022.
  • No-directional and Backward-leak Uni-directional Updatable Encryption are Equivalent, ESORICS (A conference) 2022.
  • DEKS: a Secure Cloud-based Searchable Service can Make Attackers Pay, ESORICS (A conference) 2022.
  • DEFEAT: Deep Hidden Feature Backdoor Attacks by Imperceptible Perturbation and Latent Representation Constraints, IEEE CVPR (A* conference) 2022.
  • Practical algorithm substitution attack on extractable signatures, Designs, Codes and Cryptography.
  • Two password models, IEEE ICASSP 2022.
  • PIVODL:Privacy-preserving vertical federated learning over distributed labels, IEEE Transactions on Artificial Intelligence.
  • Incrementally Updateable Honey Password Vaults, USENIX Security (A* conference) 2021.
  • Practical Threshold Multi-Factor Authentication, IEEE Transactions on Information Forensics and Security (top two computer security and cryptography publication).
  • FABRIC: Fast and Secure Unbounded Cross-System Encrypted Data Sharing in Cloud Computing, IEEE Transactions on Dependable and Secure Computing (top seven computer security and cryptography publication).
  • ROSE: Robust Searchable Encryption With Forward and Backward Security, IEEE Transactions on Information Forensics and Security (top two computer security and cryptography publication).
  • HPAKE: Honey Password-authenticated Key Exchange for Fast and Safer Online Authentication, IEEE Transactions on Information Forensics and Security (top two computer security and cryptography publication).
  • On-the-Fly Privacy for Location Histograms, IEEE Transactions on Dependable and Secure Computing (top two computer security and cryptography publication).
  • Dual Access Control for Cloud-Based Data Storage and Sharing, IEEE Transactions on Dependable and Secure Computing (top two computer security and cryptography publication).