PHD (in progress)

Title: Research and development of new Interpretable Machine Learning Models based on Compositions of Weak Base Models.

Original title in russian:

Разработка и исследование новых интерпретируемых моделей машинного обучения на основе композиций слабых базовых моделей

Masters degree (2020)

Title: Research and development of new Deep Gradient Boosting algorithms for solving high-dimensional problems.

Original title in russian:

Разработка и исследование новых алгоритмов глубокого градиентного бустинга для решения задач большой размерности

Speciality 01.04.02 – Applied Mathematics and Informatics
01.04.02_02 – Mathematical Methods for Data Analysis and Visualization
Tags machine learning, regression, classification, decision trees, gradient boosting machine, deep learning, deep forest
Annotation 
This work is devoted to development of new machine learning algorithms for regression and classification tasks based on decision trees.
The research set the following goals:
  1. Research of decision tree-based machine learning approaches.
  2. Development of new machine learning algorithms.
  3. Analysis of developed algorithms.
  4. Comparison with other decision tree-based machine learning algorithms.
The analysis of currently available decision tree-based algorithms has been made. Also, the applicability of Deep Forest to the regression task has been revised.
The study resulted in two brand new machine learning algorithms: deep gradient boosting and adaptive gradient boosting ensemble; and the modification of decision tree-based gradient boosting.
The numerical experiments made during the research, and comparison with other decision tree-based algorithms, including Random Forest, XGBoost and CatBoost, show that developed methods can be used to construct more precise models.

Bachelor degree (2018)

Title: Algorithm and software implementation of the level set method for solving the problem of semi-automatic segmentation of medical images.

Original title in russian:

Алгоритм и программная реализация метода фиксации уровня для решения задачи полуавтоматической сегментации медицинских изображений

Speciality 01.03.02 – Applied Mathematics and Informatics
01.03.02_02 – System programming
Tags image segmentation, transfer equation, medical imaging, semi-automatic segmentation, downsampling