Hongbin Zhu
Trustworthy intelligence for connected finance.
I am an Associate Professor at the Institute of Financial Technology, Fudan University. My research develops learning systems that can collaborate across institutions, reason over relational data, and operate safely in high-stakes financial environments.
Artificial Intelligence
Reliable generative and predictive AI for domain-critical financial services.
Federated Learning
Collaborative learning under privacy, heterogeneity, communication, and fairness constraints.
Graph Learning
Graph representation learning and personalized graph transformers for connected data.
Financial Security
Privacy-aware intelligence for risk, fraud, and secure financial infrastructure.
Current focus
My current work sits at the intersection of federated graph learning and financial AI: learning from distributed, scale-heterogeneous graphs without centralizing sensitive data. I am also interested in retrieval-augmented generation for trustworthy financial services and efficient edge intelligence.
Publication profile
Research spanning trustworthy AI, federated learning, graph learning, financial intelligence, edge computing, and signal processing. Explore the publication record →
Selected work
- Federated Learning with Class Imbalance Reduction, EUSIPCO, 2021.
- Online Client Selection for Asynchronous Federated Learning with Fairness Consideration, IEEE Transactions on Wireless Communications, 2022.
- DPTVAE: Data-driven Prior-based Tabular Variational Autoencoder for Credit Data Synthesizing, Expert Systems with Applications, 2024.
- PerFedGT: A Personalized Federated Graph Transformer for Scale-heterogeneous Graph Data, Information Processing & Management, 2025.
- MasterKey: A Multi-Target Backdoor Attack in Federated Learning, preprint, 2025.