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

News

Mini-workshop Stochastic Processes and Probabilistic Models in Machine Learning

On September 12 and 13, the faculty hosted a mini-workshop "Stochastic Processes and Probabilistic Models in Machine Learning" with the participation of four invited foreign specialists:

The mini workshop consisted of two blocks. Firstly, the invited specialists read lectures giving students the knowledge about parametric and nonparametric probabilistic methods application in machine learning. The second block consisted of twenty-minute reports of representatives of Russian scientific groups. Russian speakers told about particular projects where probabilistic models are used, and gave examples of how the methods explained in the lectures find application in solving various problems of machine learning.

In his lecture, Novi Quadrianto spoke about honesty and transparency in machine learning models, and how to ensure the fulfillment of these criteria. Ilya Tolstikhin acquainted listeners with a common approach for understanding two popular deep probabilistic models today: generative adversarial networks and variational autocoders. Wray Bantine explained what Dirichlet processes and Pitman-Jor processes are and described their important properties used in machine learning. Maurizio Filippone continued the topic of stochastic processes and described in detail about Gaussian processes and modern models based on them. The materials of lectures and speeches are available on the mini-workshop page.

During the visit, the invited specialists also held several scientific consultations with the faculty members on the topic of their research. In addition, Wray Bantine spoke on September 11 at the Colloquium of the Faculty of Computer Science and on September 14 read in Yandex a lecture on another type of stochastic processes - Determinantal Point Processes.