NERI Seminar Series on 19 Dec 2024 by Prof Peng Jiang: Enhancing Bayesian Networks with Hidden Variables and the Environmental Modelling Applications
We were honoured to host Associate Professor Peng Jiang from the Department of Industrial Engineering and Management, Sichuan University, for an engaging seminar as part of the NERI Seminar Series on December 19, 2024. Prof. Peng, who completed his PhD at Shanghai Jiao Tong University (SJTU) and participated in a joint PhD program at the National University of Singapore (NUS), also contributed as a Research Fellow under the E2S2-CREATE Programme (https://nus.edu.sg/neri/research/e2s2/) and a Scientist at A*STAR's Institute of High-Performance Computing before his current position at Sichuan University.
During the seminar, Prof. Peng shared his insights on "Enhancing Bayesian Networks with Hidden Variables and the Environmental Modelling Applications." His research addresses the challenges of incorporating multi-dimensional influencing variables into Bayesian networks, which often lead to unstable parameter estimates and high computational costs.
To overcome these hurdles, Prof. Peng introduced the Adaptive Hidden Factor Structured Bayesian Network (AHFSBN). This innovative approach uses a semi-supervised learning method to adaptively learn both parameters and hidden structures from multivariate spatial-temporal data.
The seminar highlighted the successful application of AHFSBN in managing risks associated with emerging aquatic contaminants, such as cyanotoxins and antimicrobial resistance genes (ARGs), in natural aquatic environments. Real-world cases demonstrated the network's potential to provide intelligent, data-driven decision support and valuable modeling insights.
Prof. Peng also discussed potential applications and the challenges of this paradigm, particularly in health risk management and the circular economy.
A big thank you to Prof. Peng for sharing his expertise and contributing to our understanding of this cutting-edge topic! A special thank you also goes to Prof. Karina Gin for chairing the seminar.
