Zgadzam się
Nasza strona zapisuje niewielkie pliki tekstowe, nazywane ciasteczkami (ang. cookies) na Twoim urządzeniu w celu lepszego dostosowania treści oraz dla celów statystycznych. Możesz wyłączyć możliwość ich zapisu, zmieniając ustawienia Twojej przeglądarki. Korzystanie z naszej strony bez zmiany ustawień oznacza zgodę na przechowywanie cookies w Twoim urządzeniu.
This work is devoted to the development of a new image classification method based on the application of a hybrid semi-supervised learning algorithm for convolutional neural networks. The proposed method is based on the active learning approach and uses metrics to identify and label samples that have a high impact on the certainty metric of the entire neighborhood. To evaluate the effectiveness of the proposed method, a synthetic semi-supervised benchmark dataset is used. The optimal choice of the graph algorithm and clustering algorithm is carried out by testing different combinations of algorithms on the synthetic dataset. The proposed method demonstrates an increase in accuracy compared to the base classifier and more efficient utilization of active labeling compared to supervised active learning. As a result, the possibility of identifying and solving the unknown class problem is realized.