Learning the Complexity of Food: From Molecules to Perception

Abstract

Food can be viewed as a high-dimensional and highly complex system, involving tens of thousands of molecules whose compositions and interactions collectively determine human perception. From a data science perspective, we frame this complexity across three layers: (1) molecular representation, describing the chemical composition of food; (2) molecular interactions, capturing the nonlinear relationships that give rise to food properties; and (3) perceptual mapping, linking molecular information to human sensory responses. Using representative case studies, this talk will illustrate how artificial intelligence—from supervised learning to generative modelling—can help learn these complex mappings and uncover structure in high-dimensional food data

Biography

Dr. Dachuan Zhang is an Assistant Professor and leads the Food Informatics and Artificial Intelligence (FoodAI) research group at NUS. He received his Ph.D. in Computational Biology from the CAS and completed postdoctoral training at ETH Zurich. As a data scientist with a deep background in food science, he focuses on developing tailored data infrastructure and data-driven approaches to unveil the complexity of our food systems. He serves as an associate editor of npj Science of Food, Nutrition & Metabolism, and the Journal of the American Oil Chemists' Society. His contributions have earned several prestigious awards, including the IUFoST Young Scientist Award, ACS Irving Sigal Global Mobility Award, Grantham Foundation Detox Award, and President’s Award of CAS.