Yee Whye Teh
Papers
2
Total Citations
228
H-Index
2
About
Yee Whye Teh is a leading figure in machine learning and Bayesian statistics, whose work has profoundly shaped modern probabilistic modeling and generative AI. His research spans Bayesian nonparametrics, Gaussian processes, and score-based generative models, with a focus on developing flexible, theoretically grounded frameworks for complex data. Teh’s seminal 2005 paper on “Semiparametric Latent Factor Models” (202 citations) introduced a novel approach using Gaussian processes to capture dependencies among multiple response variables, coupled with an efficient approximate inference scheme—a foundational contribution to latent variable modeling. More recently, his 2022 work on “Riemannian Score-Based Generative Modelling” (26 citations) advanced score-based generative models (SGMs) by incorporating geometric structure into the diffusion process, enhancing performance on data with non-Euclidean geometry. This work highlights Teh’s ongoing impact on generative AI, where his innovations in noising and generative stages have influenced cutting-edge methods. With a career marked by rigorous theoretical contributions and practical inference techniques, Teh’s research continues to inspire students and researchers, bridging Bayesian nonparametrics and modern deep learning.
Research Focus
Key Achievements
Top Papers
- 1Semiparametric Latent Factor Models202 citations · 2005
- 2Riemannian Score-Based Generative Modelling26 citations · 2022