Anh Truong

Anh Truong is a graduate student in electrical engineering and computer science combining optimization, geometry, and machine learning in service of discovering and exploiting structure in geometric data. By identifying underlying mathematical structure, his work uncovers efficiency gains through specialized algorithms that leverage mathematical properties of representations like neural fields and support functions. Anh discovered that low-rank structure naturally emerges in neural field weights during small edits, enabling 7-8x data compression through low-rank adaptation. In parallel work, he developed a regression-free method to approximate support functions of convex shapes using a variational characterization, establishing theoretical guarantees and enabling valid approximations from simple building blocks. As a MathWorks Fellow, he will expand geometry processing tools to bridge the gap between convex and non-convex geometry. Anh’s research could demonstrate how practitioners across geometric computing and beyond could achieve greater efficiency, stronger theoretical guarantees, and simpler algorithms by exploiting just the right geometric assumptions on their data.