Alexander Korotin is the head of the Generative AI research group at the Skoltech Applied AI center. His primary research field is developing scalable and theoretically justified methods to build generative models of data.  He investigates existing and proposes new generative models  based on the optimal transport & Schrodinger bridges theory which is closely related to the popular diffusion generative models. Alexander’s research results get published regularly in top ML/AI conferences and journals:  since 2020, he has presented 13 papers in A* conferences (ICLR, NeurIPS) and published several papers in Q1 journals (Neurocomputing, Pattern recognition) among others. Alexander’s research work has been highly assessed by the ML community: he has been twice  (in 2019 and 2021) awarded the Yandex ML Prize for researchers delivering cutting-edge research in ML/AI. The three most notable Alexander's research papers are "Wasserstein-2 Generative Networks" (ICLR 2021), "Neural Optimal Transport" (ICLR 2023 Spotlight, Top 25%) and "Entropic Neural Optimal Transport via Diffusion Processes" (NeurIPS 2023 Oral, Top 3%).  

In 2018 Alexander graduated from the Higher School of Economics where he studied math (bachelor) and then computer science (master). In parallel, he finished the Yandex School of Data Analysis. When a student, Alexander took an active part  in various research internships in Yandex Research, IITP RAS, Skoltech, HSE. In 2018, Alexander began his PhD studies in computer science at Skoltech under prof. Evgeny Burnaev’s supervision. In 2022, he successfully finished PhD studies, and in 2023 defended the thesis "Parametric methods for computing optimal transport maps, distances and barycenters" at FRC CSC RAS. His current research is a logical continuation of his PhD studies. At Skoltech, Alexander supervises several PhD/MSc students (~10) in their research topics related to generative modeling and optimal transport. Under his supervision, students do fundamental ML research (developing new methods for generative modeling and deriving theoretical guarantees for them) as well as apply generative models to real-world practical tasks including image translation, super-resolution, inpainting, voice style transfer, etc. For example, the recent applied project conducted by Alexander and his group at the Applied AI center is the development of neural models for super-resolution of weather maps (obtained with a climatic model) using optimal transport. These models are planned to be employed as a part of the forecasting systems for the forest fires in Russia. Beside research and student supervision, Alexander takes part in educational activities. For example, Alexander frequently gives lectures and mini courses on generative models of optimal transport. Recent presentations were given at Skoltech ML summer school (SMILES 2023) and AIRI summer schools on AI (2022, 2023). 

Selected publications (the complete list can be found in CV or Google Scholar profile): 

Paper [1] is accepted as Oral (Top 3%) to NeurIPS 2023 conference. 

Paper [2] is a Spotlight (Top 25%) paper at ICLR 2023 and is the only spotlight paper from Russia. 

[1] Gushchin, N., Kolesov, A., Korotin, A., Vetrov, D., & Burnaev, E. (2023). Entropic neural optimal transport via diffusion processes. In Thirty-seventh Conference on Neural Information Processing Systems. 

[2] Korotin, A., Selikhanovych, D., & Burnaev, E. (2023). Neural Optimal Transport. In The Eleventh International Conference on Learning Representations. 

[3] Korotin, A., Egiazarian, V., Asadulaev, A., Safin, A., & Burnaev, E. (2021). Wasserstein-2 Generative Networks. In International Conference on Learning Representations. 

Links: Google Scholar: https://scholar.google.ru/citations?user=1rIIvjAAAAAJ&hl=en 

GitHub: https://github.com/iamalexkorotin 

Website: https://akorotin.netlify.app/