THE RELATIONSHIP BETWEEN BASE CLASSIFIER DIVERSITY AND THE EFFECTIVENESS OF ENSEMBLE METHODS
Ключевые слова:
ensemble learning, classifier diversity, machine learning, CNN, LSTM, classification accuracyАннотация
An ensemble study is an effective approach to achieving high results in classification tasks, increasing accuracy and reliability compared to individual classifiers. The diversity of the main classifiers significantly affects the effectiveness of the ensemble. This study is aimed at studying the relationship between the diversity of classifiers and ensemble accuracy, and assessment was carried out based on various diversity metrics. The combinations of deep learning models such as CNN and LSTM within the ensemble were analyzed, and an attempt was made to determine how their diversity affects the effectiveness of the ensemble. Experimental results showed that a combination of classifiers with high diversity increases the accuracy of the ensemble model, especially when working with complex datasets. Based on the research results, scientific and practical recommendations were developed for the selection and combination of classifiers to increase the effectiveness of ensemble learning. This approach allows you to maximize ensemble techniques in face recognition, natural language processing, and other AI problems.
