The project is financed by the Scientific Research Fund of the MINISTRY OF EDUCATION AND SCIENCE in the competition session "Fundamental Scientific Research-2021". Funding is solely to support the implementation of non-profit scientific activity for fundamental scientific research for the acquisition of new knowledge. Non-profit scientific activity is in accordance with the National Strategy for the Development of Scientific Research in the Republic of Bulgaria 2017-2030. The project is of interdisciplinary nature, closely intertwining mathematics, statistics and computer science.
OBJECTIVES
The main objective of this project is to develop innovative mathematical and statistical machine learning (ML)-based methods and to demonstrate their advantages by applying them for intelligent data processing in the areas of environment protection, engineering and agriculture, finances and business.
The research objectives include:
development and creation of new statistical ML-based methods and algorithms with respect to information with nondynamic data and approbation of empirical data from engineering and agriculture;
development and creation of new ML-based methods, including hybrid ones, and algorithms with respect to time series and realization in the area of environment protection and finances;
creation of models of memristor-based delayed neural network of fractional order and algorithmization with appropriate numerical methods.
Key Words: Machine Learning, Ensemble and Bagging, Random Forest (RF), Classification and Regression Trees (CART), Multivariate Adaptive Regression Splines (MARS), regularized regression (RR), Memristor-based neural network
Project tasks
Development and creation of new statistical methods and algorithms with MO for classifications in multidimensional data.
New theoretical results to create methods for discovering new knowledge about dependencies in data difficult to process with classical methods.
New results and algorithms by creating a hybrid type of methods.
Theoretical and numerical study of different types of stability of memristor-based neural networks of fractional order with delay