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

About the Laboratory

Currently the world is witnessing a revolution in machine learning on big data associated with adoption of deep neural networks and deep learning. In the years 2010-2014, a number of problems were solved by means of neural networks. The more complex problems require building probabilistic models with neural networks as one of the components. The standard framework for such models is the Bayesian approach. The results published in 2013-2016 allow using this approach for big data. Starting from 2015, an evident trend is the merging of deep learning and Bayesian methods. Neural networks are utilized for variational inference in complex probabilistic models providing scalability to big data. On the other hand, Bayesian regularization allows to create compact automatically tuned neural structures with high generalization ability. There are almost no scientific groups working in the aforementioned areas in Russia. One of the exceptions is the Bayesian Methods Research Group led by Prof. Dmitry Vetrov. The laboratory is established on the basis of this group.

Dr. Novi Quadrianto is the leading international researcher of the laboratory. He is an assistant professor in the University of Sussex. He is an expert in Gaussian processes and their application to multisource data analysis. The last results in Bayesian regularization in neural networks allow to interpret this type of regularization as a scalable generalization of Gaussian processes. This fact enables to join the experience of Bayesian Methods Research Group and Dr. Novi Quadrianto towards fruitful joint research.


 

Have you spotted a typo?
Highlight it, click Ctrl+Enter and send us a message. Thank you for your help!