Laboratory of Structural Methods of Data Analysis in Predictive
Modeling Moscow Institute of Physics and Technology
ENG
Логин:
Пароль:

Направления исследований

Statistical methods
The use of statistical methods for analyzing data has become common practice in virtually all scientific disciplines. Classical statistical methods are related to maximum likelihood and Bayesian approaches. These approaches are usually based on parametric or non-parametric probabilistic data models and are cornerstone in many applications developed at PreMoLab. They include, in particular, inverse problems, smoothing techniques, surrogate modeling, time series and optimal stopping.

подробнее »

Structural adaptive inference
Structural adaptive inference provide basis for most of data analysis algorithms developed at PreMoLab. This approach enables efficient analysis of complex statistical models.

подробнее »

Structural optimization
Application-specific and structure-specific optimization methods; optimization looking “inside the black box.”

подробнее »

Huge-scale problems
Description will be provided later.

подробнее »

Primal-dual subgradient methods
Primal-dual subgradient schemes for nonsmooth convex optimization problems

подробнее »