• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Bayesian Methods Research Group

News

4

29 May 2018

Samsung has officially opened new Samsung AI Center in Moscow

We're excited to announce that Samsung has officially opened new Samsung AI Center in Moscow, where among others professor Vetrov and our group will be involved in research on Bayesian Methods and Deep Learning.

26 April 2018

The paper on Conditional Generators of Words Definitions has been accepted to ACL 2018!

The work proposes conditional language model for word’s definitions and shows that one particular word embedding contains information about most meanings of the corresponding word.

Bayesian methods research group is located in two places: Higher School of Economics, Moscow, and d Computational Mathematics and Cybernetics Department of Moscow State University. It approximately consists of 3 postgraduates (including one Ph.D.), 5 Ph.D. students, and 15 students. In 2017 we established new International Lab of Deep Learning and Bayesian Methods at the faculty of computer sciences in Higher school of economics. We carry out research work on the development of new machine learning and Bayesian inference algorithms which take into account the specific features of given problem. We widely exploit Bayesian framework and the theory of graphical models in particular.

Recent years have proved that the more data is involved into analysis, the better (often much better) results one may obtain. The breakthrough in machine learning has happened due to the successful application of deep neural networks which turned to be extremely powerful when dealing with huge amounts of data. However, it is now clear that classical methods simply do not work when one needs to process extremely large datasets. So New Mathematics or the mathematics of Big Data Age is needed. Our group is involved in the process of developing such mathematics and is carrying out research in deep learning, stochastic optimization, tensor decompositions, scalable variational inference. Our efforts are supported by YandexNVidiaKaspersky lab, Samsung, Sberbank, SchlumbergerJetBrains.

Important directions of our work are applied projects from many domains including text processing, computer vision, software code analysis. During the work over the projects, the students gain practical experience of using different algorithms from computer science as well as software engineering skills. We strongly encourage the research activity of students and the publication of papers authored or co-authored by students. 

Our group is involved in teaching process in Higher School of Economics, Yandex School of Data Analysis and Moscow State University.