Journal Articles - Computer and Information Technology - 2021

Permanent URI for this collection

http://repository.vlu.edu.vn:443/handle/123456789/102

Browse

Browsing Journal Articles - Computer and Information Technology - 2021 by Author "Tai Vovan"
  • Publication
    Automatic clustering algorithm for interval data based on overlap distance
    ( 2021)
    Ngoc Lethikim
    ;
    Tuan Lehoang
    ;
    Tai Vovan
    In this study, the improved overlap distance is used as a criterion in order to build clusters for interval data. This distance has shown the suitability, and given an outstanding advantage in evaluating the similarity for intervals with a lot of the considered data sets. Based on the overlap distance, we propose the Automatic Clustering Algorithm for Interval data (ACAI). One of the best advantages of the proposed algorithm is that ACAI figure out simultaneously the appropriate number of groups, and factors in every group. The proposed algorithm can be effectively performed through a Matlab procedure. Based on the extracted intervals from texture of images, we have applied ACAI to recognize the images, an interesting and challenging issue at present. Experimental data sets including the differences of the characteristics as well as the number of elements has shown the reasonableness of the proposed algorithm, and its advantages in comparing to the surviving ones. From the image recognition problem, this research has shown prospect in practical applications for many fields.
  • Publication
    Fuzzy clustering algorithm for outlier-interval data based on the robust exponent distance
    ( 2021)
    Dinh Phamtoan
    ;
    Khanh Nguyenhuu
    ;
    Tai Vovan
    The outlier elements of a data are ones that differs significantly from others. For many reasons, we have to face with outlier elements in data analysis for the different fields. Because an outlier element can cause the serious problems in statistical analyses, studying about it is interested in many researchers. This article proposes the fuzzy clustering algorithm for outlier - interval data based on the robust exponent distance to overcome the drawback of traditional clustering algorithm which to clean the outliers before performing. The outstanding advantage of this algorithm is to find the suitable number of clusters, to cluster for the interval data with outlier elements, and to determine the probability belonging to clusters for the intervals at the same time. The proposed algorithm is described step by step via numerical examples, and can be performed effectively by the Matlab procedure. In addition, it also applied in reality with the air pollution, mushroom, and image data sets. These real applications demonstrate the robustness of the proposed algorithm in comparison with the existing ones.