نشریه مهندسی برق و مهندسی کامپیوتر ایران
سال هفدهم شماره 3 (پیاپی 52، پاییز 1398)

A robust classification model is a non-standard model for classifying learning based on an uncertain data set. An incautious model is said to have any meaningless answer to any classification model in its possible set of possible solutions. The optimal answer for a cautious robust classification model for a training data set may not be the hyper-page, in which case it will not be possible to classify the data at the test stage. In this paper, incautious robust classification models are introduced and their problems are investigated and then by changing the loss function of a robust classifier, a cautious robust classification model is presented to prevent incautious. The proposed cautious model is standardized and solutions are provided to reduce the training time and test time. In the experiments, the proposed model was compared with some incautious robust models to classification incomplete training data set, and complete definitive training data set. The results showed that in the incomplete data set, the proposed model had less training time and error rate than incautious models. Also, in the complete definitive data set, the proposed model training time and test time were less than incautious models. The results approved that adding caution to a robust classifier is efficient.

Fast development in the utilization of cloud computing leads to publishing more cloud services on the cloud environment. The single and simple services cannot satisfy the users’ real-world complex requirements. To create a complex service, it is necessary to select and compose a set of simple services. Therefore, it is essential to embed a service composition system in cloud computing environment. Service composition is one of the important NP-hard problems in the service-oriented computings. In this paper, a biogeography-based optimization algorithm is used to create the optimal composite-services. The proposed method was simulated and executed on five different scenarios with different number of tasks and candidate services. The throughput of the proposed method, genetic algorithm and particle swarm optimization algorithm are respectively 0.9997, 0.9975 and 0.9994; furthermore, the reliability of these methods are respectively 0.9993, 0.9980 and 0.9982. The results of simulations indicate that the proposed method outperforms the previous methods in the same conditions in terms of throughput, successability, reliability, response time, and stability.

The main goal of web search engines is to find the most relevant results with respect to the user query in a shortest possible time. To do so, the crawled documents have to be partitioned between several servers in order to use their aggregate retrieval and processing power. The search engines use different policies for efficient partitioning of documents. In this paper, we propose a new document partitioning method that intends to balance the load between servers to reduce the response time of queries. The idea is to weigh each term based on its daily frequency in log of past queries. We then assign a weight to each document via summing the weight of its substituent terms. The weight of a document approximates the likelihood of its presence in future search results. Finally, the documents are partitioned between servers in a way that the sum of document weights in each server becomes roughly equal. Our evaluation results show that the proposed method is able to balance the load by about 20% better than former algorithms, especially in the peak of search engine traffic.

Distributed processing uses local computations at each node and communications among neighboring nodes to solve the problems over the entire network. Diffusion is one of the methods for performing distributed networks. This paper presents a novel Variable Step-Size Diffusion Affine Projection Algorithm (VSS-DAPA) to improve the performance of the Diffusion Affine Projection Algorithm (DAPA) in distributed networks. The variable step-size of each node is obtained by minimizing the Mean-Square Deviation (MSD) in that node. In comparison with Diffusion Affine Projection Algorithm (DAPA), the VSS-DAPA algorithm has faster convergence speed and lower steady-state error. To reduce the computational complexity of VSS-DAPA, the Variable Step-Size Selective Regressors Diffusion Affine Projection Algorithm (VSS-SR-DAPA), the Variable Step-Size Dynamic Selection of Diffusion Affine Projection Algorithm (VSS-DS-DAPA) and Variable Step-Size Selective Partial Update Diffusion Affine Projection Algorithm (VSS-SPU-DAPA) are proposed. Simulation results show the good performance of proposed algorithms in convergence speed and steady-state error.

Wireless sensor networks, as an up-to-date technology, are one of the fastest growing technologies in the world today. Since these networks are used in military and agricultural environments as well as for observation of inaccessible environments, these networks need to be organized to achieve goals such as successful and timely sending of data to the main station. Clustering of wireless sensor networks is one of the most widely used methods for organizing these networks. Various ways to cluster these networks are provided, most of which are aimed at preventing energy loss and increasing the lifetime of sensor nodes. The thesis attempts to present a new geometric method for clustering the nodes of wireless sensor networks. In this geometric method, Apollonius circle is used to draw the abstract shape of the clusters and to assemble the nodes around the cluster head. Due to the high accuracy that it has in determining the fit of node distances, this circle can accurately assign nodes to cluster heads and prevent large single-node clusters or faraway nodes. In this algorithm, a main station, a number of nodes are used as a cluster header and a number of nodes as routers. The goal is to find the most accurate cluster heads and create clusters of high coverage in the network. The proposed method is implemented in MATLAB software and comparison of the results obtained from the view of successful data transmission, clustering accuracy, network lifetime and number of coverage areas, is showing accuracy of this method compared to optimal Leach algorithms and K-means presented in this field.

The Internet of Things (IOT) is a new concept in the area of monitoring information transmission and remote control of things, existents and equipment that has been able to adapt itself with different industries and substructures easily. The information transmission with regard to the non-homogenous environment of internet of things has been a challengeable topic and use of routing methods by considering the limitations of processing, calculating, saving and communicating has been known as a necessary issue. Various algorithms with special applications have been already introduced in the domain of internet of things and wireless sensor networks that each one somehow has been successful in achieving the routing goals. Some proposed protocols in this field have used a tree structure for gathering the network information. These methods in selecting the parent or children of graph are affected by important challenges depending on the type of application. In this paper, at first, a general classification of advantages and defects of recent methods has been done in the domain of routing the internet of things and then a routing method service quality awareness in routing based on fuzzy system has been suggested. The results of simulations and assessment express that the suggested method in the tests of energy productivity, delay ratio and data delivery ratio have better performance than the recent methods.

Due to the explosive increment of data in recent application, available data storage cannot respond to this volume of data, for this reason molecular memory have been suggested in recent research. DNA is molecular data storage that can store a large amount of data in a limited space with high endurance. Storing data in low volumes can be provided using appropriate mapping. In this paper, a new method for mapping digital data to DNA have been proposed with the aim of simple coding, omitting the decoding faults, increasing the speed of coding and storing digital data and sign-digit with sufficient compression. Studies show that the proposed method can guarantee long-term retrieval of information from DNA compared to previous methods. It also uses less compression to store digital data as compared to the previous methods.

The methods used today to investigate brain connections to diagnose brain-related diseases are the imaging method of resting magnetic resonance imaging. In this paper, a new method is proposed using an evolutionary variable-length algorithm to select the appropriate features to improve the accuracy of the diagnosis of healthy and patient-to-patients with attention deficit hyperactivity disorder based on analysis of rs-fMRI images. The characteristics examined are the correlation values between the time series signals of different regions of the brain. Selection of the variable-length property were based on the honey bee algorithm in order to overcome the problem of feature selection in algorithms with fixed-length vector lengths. The Mahalanubis distance has been used as a bee algorithm evaluation function. The efficiency of the algorithm was evaluated in terms of the value of the evaluation function in the first degree and the processing time in the second degree. The results obtained from the significantly higher efficiency of the variable-length bee algorithm than other methods for selecting the feature. While the best result of the overall categorization accuracy among the other methods with the 26 selected characteristics of the PSO algorithm is 76.61%, the proposed method can achieve a total classification accuracy of 85.32% by selecting 25 features. The nature of the data is such that the increase in the number of attributes leads to a greater improvement in the accuracy of the classification so that by increasing the length of the characteristic vector to 35 and 45, classification accuracy was 91.66% and 95.57% respectively.