Skoltech presented at the largest international conference ICML

Researchers from the AI and Applied AI centers joined the International Conference on Machine Learning 2024 in Vienna, one of the three premier A* international conferences on machine learning and artificial intelligence, alongside NeurIPS and ICLR.

On July 23, Assistant Professor Ekaterina Muravleva and Professor Ivan Oseledets presented a poster titled “Neural Operators Meet Conjugate Gradients: The FCG-NO Method for Efficient PDE Solving,” introducing the FCG-NO method that leverages neural operators to enhance the accuracy of differential equation solutions. 

On July 24, Research Engineers Petr Mokrov and Milena Gazdieva, Professor Evgeny Burnaev, and Assistant Professor Alexander Korotin discussed their research titled “Estimating Barycenters of Distributions with Neural Optimal Transport,” which proposes a novel method for finding barycenters of distributions based on neural networks and optimal transport. Professor Evgeny Burnaev, PhD student Nikita Balabin, and Leading Research Scientist Sergey Barannikov presented a paper titled “Disentanglement Learning via Topology,” where they developed the TopDis method for learning disentangled data representations using topological losses. Head of Development Team Sergey Shumilin, Professor Evgeny Burnaev, and Head of Research Group Vladimir Vanovskiy showcased their poster under the title “Self-Supervised Coarsening of Unstructured Grid with Automatic Differentiation,” presenting a method for coarsening computational unstructured grids using automatic differentiation in a self-supervised mode. 

On July 25, Research Engineer Nikita Gushchin, Junior Research Engineer Sergey Kholkin, Professor Evgeny Burnaev, and Assistant Professor Alexander Korotin presented their study titled “Light and Optimal Schrödinger Bridge Matching,” featuring their LightSB-M method for optimally solving the Schrödinger bridge problem, which improves the efficiency of diffusion process modeling. 

Co-authors of the works included in the ICML program from Skoltech also include PhD students Darya Voronkova, Alexander Kolesov, Denis Kuznedelev, Junior Research Scientist Alexander Rudikov, Research Engineers Alexander Ryabov and Igor Udovichenko, Head of Research Group Ilya Trofimov, Research Scientist Vladimir Fanaskov, and Senior Research Scientist Nikolai Yavich.