Pattern Analysis and Intelligent Systems
Ehsan Soleimani; Kamal Mirzaie
Articles in Press, Accepted Manuscript, Available Online from 30 July 2022
Abstract
With the increasing use of online services in the form of platforms or websites, as well as the growth of users in online spaces, it seems necessary to study and analyze these networks. One of the areas of interest for social media analysts is Community Detection (CD) in these networks. The purpose of ...
Read More
With the increasing use of online services in the form of platforms or websites, as well as the growth of users in online spaces, it seems necessary to study and analyze these networks. One of the areas of interest for social media analysts is Community Detection (CD) in these networks. The purpose of CD is to discover subcommunities within the original networks. Identifying these groups in social networks has many applications in discovering new groups, and extracting common points between users. The main challenge in CD is to find the relationships between the nodes and form clusters in them. In particular, the use of CD techniques makes it possible to identify similar nodes that have common features. Because user data does not belong to a specific category and there is a lot of data in the community and a similar relationship should be established between them based on the type of data. In this paper, a new method for CD using an Adaptive Genetic Algorithm (AGA) is proposed. The evaluation of the proposed model was performed on the Enron Email suite using Normalized Mutual Information (NMI), Variation Information (VI) and Cross Common Fraction (CCF) criteria. The results show that with increasing the number of generations, the accuracy of the proposed model increases and also the rate of crossover and mutation are very effective in the accuracy of detection. The accuracy of NMI, VI and CCF criteria with 800 iterations is 82.17, 72.95 and 75.16, respectively.
Computer Networks and Distributed Systems
Mohsen Raji; Saeed Keshavarzi; Mehran Farvardin
Articles in Press, Accepted Manuscript, Available Online from 07 November 2022
Abstract
A Wireless Sensor Network (WSN) is consisted of thousands of sensor nodes to monitor environmental conditions. Energy plays a major role in these networks because nodes are worked with batteries and are sometimes placed in an inconsistent environment where their battery cannot be charged or replaced. ...
Read More
A Wireless Sensor Network (WSN) is consisted of thousands of sensor nodes to monitor environmental conditions. Energy plays a major role in these networks because nodes are worked with batteries and are sometimes placed in an inconsistent environment where their battery cannot be charged or replaced. In this paper, a new dynamic clustering method is presented for WSNs with moving targets. In this method, target tracking is carried out using the speed and direction of target movement as well as selection of the optimal cluster-head (CH) using the artificial bee colony (ABC) meta-heuristic algorithm with fuzzy TOPSIS. In the proposed CH selection method, three criteria are considered; i.e. the remaining energy of the node, the cost of selecting each node as the CH, and the risk of hardware failure of each node. Since each of these criteria should be assigned with a suitable weight, ABC optimization algorithm is used to find the best weights for ranking decisions. After specifying the objective function in the ABC algorithm and weighting the criteria, fuzzy TOPSIS algorithm is used and each of the nodes is fuzzified in each of the criteria that are normalized and weighted. The results show that in addition to maintaining the accuracy of tracking the moving target, the proposed method achieves a 5.4% improvement in network energy consumption in comparison with a state-of-the-art method called EEAC.
Pattern Analysis and Intelligent Systems
Farhad Soleimanian Gharehchopogh; Berivan Rostampnah
Volume 7, Issue 3 , August 2021, , Pages 177-186
Abstract
Abstract— Clustering is one of the most popular techniques in unsupervised learning in which data is divided into different groups without any prior knowledge, and for this reason, clustering is used in various applications today. One of the most popular algorithms in the field of clustering is ...
Read More
Abstract— Clustering is one of the most popular techniques in unsupervised learning in which data is divided into different groups without any prior knowledge, and for this reason, clustering is used in various applications today. One of the most popular algorithms in the field of clustering is the k-means clustering algorithm. The most critical weakness of k-means clustering is that it is sensitive to initial values for parameterization and may stop at local minima. Despite its many advantages, such as high speed and ease of implementation due to its dependence on the initial parameters, this algorithm is in the optimal local configuration and does not always produce the optimal answer for clustering. Therefore, this paper proposes a new model using the Bald Eagle Search (BES) Algorithm with the Sine Cosine Algorithm (SCA) for clustering. The evaluation of the proposed model is based on the number of iterations, convergence, number of generations, and execution time on 8 UCI datasets. The proposed model is compared with Flower Pollination Algorithm (FPA), Crow Search Algorithm (CSA), Particle Swarm Optimization (PSO), and Sine-Cosine Algorithm (SCA). The results show that the proposed model has a better fit compared to other algorithms. According to the analysis, it can be claimed that the proposed model is about 10.26% superior to other algorithms and also has an extraordinary advantage over k-means.
Software Engineering and Information Systems
Mohammad Reza Hassanzadeh; farshid keynia
Volume 7, Issue 1 , February 2021, , Pages 35-54
Abstract
Metaheuristic algorithms are typically population-based random search techniques. The general framework of a metaheuristic algorithm consisting of its main parts. The sections of a metaheuristic algorithm include setting algorithm parameters, population initialization, global search section, local search ...
Read More
Metaheuristic algorithms are typically population-based random search techniques. The general framework of a metaheuristic algorithm consisting of its main parts. The sections of a metaheuristic algorithm include setting algorithm parameters, population initialization, global search section, local search section, and checking the stopping conditions in a metaheuristic algorithm. In the parameters setting section, the user can monitor the performance of the metaheuristic algorithm and improve its performance according to the problem under consideration. In this study, an overview of the concepts, classifications, and different methods of population initialization in metaheuristic algorithms discussed in recent literature will be provided. Population initialization is a basic and common step between all metaheuristic algorithms. Therefore, in this study, an attempt has been made that the performance, methods, mechanisms, and categories of population initialization in metaheuristic algorithms. Also, the relationship between population initialization and other important parameters in performance and efficiency of metaheuristic algorithms such as search space size, population size, the maximum number of iteration, etc., which are mentioned and considered in the literature, are collected and presented in a regular format.
Pattern Analysis and Intelligent Systems
Farhad Soleimanian Gharehchopogh; Sevda Haggi
Volume 6, Issue 2 , May 2020, , Pages 79-90
Abstract
The detection and prevention of crime, in the past few decades, required several years of research and analysis. However, today, thanks to smart systems based on data mining techniques, it is possible to detect and prevent crime in a considerably less time. Classification and clustering-based smart techniques ...
Read More
The detection and prevention of crime, in the past few decades, required several years of research and analysis. However, today, thanks to smart systems based on data mining techniques, it is possible to detect and prevent crime in a considerably less time. Classification and clustering-based smart techniques can classify and cluster the crime-related samples. The most important factor in the clustering technique is to find the centrality of the clusters and the distance between the samples of each cluster and the center of the cluster. The problem with clustering techniques, such as k-modes, is the failure to precisely detect the centrality of clusters. Therefore, in this paper, Elephant Herding Optimization (EHO) Algorithm and k-modes are used for clustering and detecting the crime by means of detecting the similarity of crime with each other. The proposed model consists of two basic steps: First, the cluster centrality should be detected for optimized clustering; in this regard, the EHO Algorithm is used. Second, k-modes are used to find the clusters of crimes with close similarity criteria based on distance. The proposed model was evaluated on the Community and Crime dataset consisting of 1994 samples with 128 characteristics. The results showed that purity accuracy of the proposed model is equal to 91.45% for 400 replicates.
Computer Networks and Distributed Systems
Zoleikha Azizi; Kambiz Majidzadeh
Volume 3, Issue 4 , November 2017, , Pages 195-202
Abstract
Considering the great significant role that routing protocols play in transfer rate and choosing the optimum path for exchange of data packages, and further in the amount of consumed energy in the routing protocol, the present study has focused on developing an efficient compound energy algorithm based ...
Read More
Considering the great significant role that routing protocols play in transfer rate and choosing the optimum path for exchange of data packages, and further in the amount of consumed energy in the routing protocol, the present study has focused on developing an efficient compound energy algorithm based on cluster structure which is called active node with cluster structure. The purpose of this algorithm is to distribute the heavy traffic of data and equal load of highly-consumed energy throughout the networks by introducing unequal and unbalanced clusters into the network. In the second stage, the light sensor mechanism which is called active node sensor algorithm has been proposed. The major purpose of this mechanism is to prevent excessive interfering data of sensors through incorporating a set of active nodes in each cluster with a defensive shield near to the incident node. The third stage has aimed at proposing an active node algorithm for complexity of internal and external addressing due to clusters routing in high density distribution based on the values within node range. The obtained results indicate relative success of the design in terms of energy optimization on the basis of routing protocol.