Leo Klarner

Papers

Total Citations

H-Index

About

Leo Klarner is a rising force in generative AI, whose work bridges the theoretical and the applied. His research focuses on the cutting-edge intersection of **diffusion models** and **constrained domains**, specifically extending these powerful generative algorithms to complex, non-Euclidean spaces like Riemannian manifolds. Klarner’s major contribution lies in tackling the fundamental challenge of ensuring that generated samples adhere to physical or geometric constraints—a critical need for applications in the natural sciences. His 2023 paper, “Diffusion Models for Constrained Domains,” has already garnered **5 citations** for laying the groundwork in this nascent field. He further advanced the methodology with “Metropolis Sampling for Constrained Diffusion Models” (**3 citations**), introducing a novel sampling technique that dramatically improves the quality and validity of generated data on manifolds. This work is particularly notable for its potential to revolutionize molecular design and computational chemistry, where generating valid, constrained molecular structures is paramount. Klarner’s research is rapidly establishing him as a key innovator in making generative AI not just powerful, but also physically plausible.