Scientific activity

The Centre conducts research at the intersection of two rapidly developing fields of data analysis today: deep learning and Bayesian machine learning methods. Deep learning is a field that involves building very complex models (neural networks) to solve tasks such as image or music classification, transferring artistic style from a painting to a photograph, or predicting the next words in a text. Within the Bayesian approach, probabilistic models based on the apparatus of probability theory and mathematical statistics are considered for solving such problems. In recent years, Bayesian methods have been actively used in combination with neural network models to enhance their predictive ability, facilitate the process of tuning them to specific data, and to solve a wider range of problems. Scientific papers on this topic constitute a significant section of leading world machine learning conferences (NeurIPS, ICML, ICLR). Centre staff regularly publish at these conferences and participate in the workshops held at them.

Centre's Research Directions:

• Approximate Bayesian Inference
• Variational Autoencoders
• Tensor Neural Networks
• Acceleration of Convolutional Networks
• Stochastic Optimization Methods
• Diffusion Models.

The group regularly holds a research seminar discussing current problems in artificial intelligence and machine learning research, including the results of their own scientific work.

📆 Current schedule and archive of the special seminar's activities for 2024-2026

🎥 BayesGroup YouTube channel: recordings of special seminars for 2024-2026

🗃️ Archive of the special seminar's activities for 2016-2023

📼 Archive of video recordings of the special seminar for 2017-2023