REVIEW ON MODERN ENSEMBLE TECHNIQUES IN MACHINE LEARNING TO IMPROVE THE QUALITY OF CLASSIFICATION

Kubrakova Ekaterina Alexandrovna

The article examines the features of using ensemble techniques in machine learning to improve the quality of classification. The main scientific and practical studies on software implementation, construction of models and algorithms of ensembling for use in various fields are highlighted. Relevance. Data mining, used in machine learning, often faces various tasks, among which one of the key ones is the implementation of classification. It involves dividing the data into certain categories according to predefined classes. The ensemble method usually uses a combination of several classifiers. This is done in order to improve the results compared to the output data provided by each classifier individually. The key idea is that by combining responses in a variety of ways, individual errors can be eliminated, thereby achieving a higher overall quality of the solution within the ensemble. The purpose of the article is to highlight the main aspects of the application of ensemble techniques in machine learning to improve the quality of classification. The result of the research is the study of scientific, theoretical, practical provisions, the identification of the main approaches to the construction of ensembles of algorithms. In conclusion, the author's conclusions are given.

Keywords:ensemble, algorithm, classification task, boosting, bagging, stacking.