Method for Efficiency Increasing of Distributed Classification of the Images based on the Proactive Parallel Computing Approach

In this paper we propose a new clustering method based on alpha-procedure that differs with proactive computing of graphs and other possible factors superposition of classification. This approach does not increase the calculation time but obtains a potentially more accurate result for object classif...

Full description

Saved in:
Bibliographic Details
Published inAdvances in electrical and computer engineering Vol. 18; no. 2; pp. 117 - 122
Main Authors MUKHIN, V., VOLOKYTA, A., HERIATOVYCH, Y., REHIDA, P.
Format Journal Article
LanguageEnglish
Published Suceava Stefan cel Mare University of Suceava 01.01.2018
Subjects
Algorithms
Classification
Clustering
Computer networks
distributed computing
Distributed processing
Efficiency
Image classification
Mathematical problems
Methods
mobile agents
Parallel processing
parallel programming
Pattern recognition
performance analysis
Online AccessGet full text
ISSN1582-7445
1844-7600
DOI10.4316/AECE.2018.02015

Cover

More Information
Summary:In this paper we propose a new clustering method based on alpha-procedure that differs with proactive computing of graphs and other possible factors superposition of classification. This approach does not increase the calculation time but obtains a potentially more accurate result for object classification. The description of the method alpha-procedure parallelization is performed. The paper presents a method of parallel computing of the independent parts in distributed systems using a model of actors. The paper describes a modified method of classification of multidimensional objects designed for computing on distributed computer systems. It is shown, that in case if the number of nodes is increased, the performance decreases linearly. This method allows realizing the choosing by n-factors despite the classical method. This change is acceptable as the ways of the calculations are independent from each other and can be calculated in parallel on a distributed system. Therefore, in modern distributed systems, in case if the amount of resources is sufficient, we may use the redundant calculations that may improve the clustering results. For this method, we propose to use the resources for parallel calculation in several independent ways, and further comparing them, to choose the best result.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1582-7445
1844-7600
DOI:10.4316/AECE.2018.02015