neural network

Open Access Article

1 - Solving Multi-Criteria Decision Making Problems Using Artificial Neural Networks
M. abdoos N. Mozayani

20.1001.1.16823745.1386.5.9.7.2

Decision making is finding the best compromised solution from all feasible alternatives. Multi-criteria decision making is one of the most applied branches of decision making. Many methods have been presented for solving MCDM problems ever since. Among these methods, si More

Decision making is finding the best compromised solution from all feasible alternatives. Multi-criteria decision making is one of the most applied branches of decision making. Many methods have been presented for solving MCDM problems ever since. Among these methods, simple additive weighting, SAW, is the most commonly used method. In this paper, two methods are proposed for solving MCDM problems based on artificial neural networks. This paper shows an application of soft computing techniques in classic problems, such as decision making. Herein, two methods are presented based on both supervised and unsupervised neural networks. The results of the methods have been compared with SAW. Manuscript profile

Open Access Article

2 - An Adaptive Wavelet-Based Signal Denoising Schem
M. nasri H. Nezamabadi-pour S. Saryazdi

20.1001.1.16823745.1387.6.13.2.2

In this paper, a new class of nonlinear thresholding functions with a tunable shape parameter for wavelet-based signal denoising is presented. In addition, a new learning technique for training of thresholding neural network is introduced. Unlike to existing methods, bo More

In this paper, a new class of nonlinear thresholding functions with a tunable shape parameter for wavelet-based signal denoising is presented. In addition, a new learning technique for training of thresholding neural network is introduced. Unlike to existing methods, both the shape and the threshold parameters are tuned simultaneously using LMS rule. This permits us to consider the effects of both the threshold and the shape parameters on denoising. The proposed functions are tested in both universal-threshold and subband-adaptive denoising and compared with conventional functions. In addition, to evaluate the proposed training method, several numerical examples are performed. The experimental results obtained from denoising of several standard benchmark signals confirm the efficiency and effectiveness of the proposed methods. Manuscript profile

Open Access Article

3 - Determining of Classifiers Behavior Using Hidden Markov Model Based Decision Template
H. Sadoghi Yazdi

20.1001.1.16823745.1387.6.14.10.2

Studying of classifier behavior is interested from viewpoint of error checking and presentation of suitable solution for decreasing error rates and decreasing performance. Weakness operation of recognition system is because of small number of training samples, noisy sam More

Studying of classifier behavior is interested from viewpoint of error checking and presentation of suitable solution for decreasing error rates and decreasing performance. Weakness operation of recognition system is because of small number of training samples, noisy samples, unsuitable extracted features, method of determining of system response. Presentation of suitable model for behavior or response of recognition system, we can improve operation of recognition system. In this paper, a new hidden Markov model based decision template is generated for modeling of neurons behavior in neural network. In existing methods, relation of neurons and interaction between them is not studied whereas; response of neural network includes response value of all neurons. So, relations of neurons are modeled using new hidden Markov decision templates. This method is used into three applications include recognition of Farsi number images, normal traffic in internet network, and recognition of types of vehicles. Increasing performance of neural network indicates to superiority of the proposed system. Manuscript profile

Open Access Article

4 - Application of Neuro Space Mapping in Modeling Semiconductor Devices
M. Gordi Armaki S. E. Hosseini Mohammad Kazem Anvarifard

20.1001.1.16823745.1389.8.2.7.9

In this paper an efficient method for modeling semiconductor devices using the drift-diffusion (DD) model and neural network is presented. Unlike HD model which is complicated, time consuming with high processing cost, the proposed method has lower complexity and higher More

In this paper an efficient method for modeling semiconductor devices using the drift-diffusion (DD) model and neural network is presented. Unlike HD model which is complicated, time consuming with high processing cost, the proposed method has lower complexity and higher simulate speed. In our method, a RBF neural network is used to modify DD parameters. The modified DD model can generate simulate results of accurate HD model. The proposed method is first applied to a silicon n-i-n diode in one dimension, and then to a silicon thin-film MOSFET in two dimensions, both for interpolation and extrapolation. The obtained results for basic variables, i.e., electron and potential distribution for different voltages, confirm the high efficiency of the proposed method. Manuscript profile

Open Access Article

5 - Training of MLP Neural Network for Data Classification by GSA Method
M. Dehbashian Seyed-Hamid Zahiri

20.1001.1.16823745.1389.8.4.5.1

Nowadays, several techniques have presented for data classification. One of these techniques is neural network that has attracted many interests. In this classifier, selection a suitable learning method is very important for training of the network. Error back propagati More

Nowadays, several techniques have presented for data classification. One of these techniques is neural network that has attracted many interests. In this classifier, selection a suitable learning method is very important for training of the network. Error back propagation is the most usual training method of neural networks that late convergence and stopping in local optimum points are its weakness. New approach in neural networks training is the usage of heuristic algorithms. This paper suggests a new learning method namely gravitational search algorithm (GSA) in training of neural network for data classification. GSA method is the latest and the most novel version of swarm intelligence optimization methods. This algorithm is inspired fby the law of Newtonian gravity and mass concept in nature. In this paper, a MLP neural network is trained for classification of five benchmark data set by GSA method. Also, the proposed method efficiency in training and testing of neural network compared with those of two training methods error back propagation and particle swarm optimization. Final results showed the GSA method extraordinary performance for data correct classification in most of cases. Also, in these experiments the GSA method produced stable results in all of cases. In addition, the run time of GSA method is shorter than that of the PSO. Manuscript profile

Open Access Article

6 - Neural Control of the Induction Motor Drive: Robust Against Rotor and Stator Resistances Variations and Suitable for Very Low and High Speeds
H. Moayedi Rad M. A. Shamsi-Nejad mohsen Farshad

20.1001.1.16823745.1390.9.2.7.6

In this paper, induction motor speed control drive is designed with application two multilayer feed-forward neural networks. That those are used one for generate PWM pulse and other for estimation of required torque and flux information. For trained of the PWM wave gene More

In this paper, induction motor speed control drive is designed with application two multilayer feed-forward neural networks. That those are used one for generate PWM pulse and other for estimation of required torque and flux information. For trained of the PWM wave generate neural network is used from compound information two voltage and current classic model. Also, against general classic models for generate of the switching pulses is used as compound from reference voltage and current two motor phases. With these ideas are eliminated problems of the voltage and current classic models (flux saturation in current model for high speeds and voltage drop in voltage model for low speeds). As voltage profile is improved in this paper. The required feedback signals estimation (including: rotor flux, torque, etc.) is estimated by multilayer feed-forward neural network. That for robustness of the above estimator against rotor and stator resistances variations in time work of motor is used from compound trained data of the voltage and current classic models, because the voltage and current of the general classic models to sequence are independent of rotor and stator resistances. The simulation results by MATLAB-Simulink verify the proposed drive in improvement of the speed profile in transient and steady-state operating modes. Also, it verify clearly robust of the proposed drive against rotor and stator resistances variations in time work. Manuscript profile

Open Access Article

7 - Modeling and Analysis Iterated Prison Dilemma Game by Grossberg Counter-Propagation Neural Network
Gh. A. Montazer N. Rastegar Ramshe Alireza Askarzadeh

20.1001.1.16823745.1391.10.2.5.1

Most of the time effective decisions in strategic situations such as competitive issues require a non-linear mapping between stimulus and response. Artificial neural networks can be an appropriate way for modeling and solving these kinds of problems. Prison Dilemma Game More

Most of the time effective decisions in strategic situations such as competitive issues require a non-linear mapping between stimulus and response. Artificial neural networks can be an appropriate way for modeling and solving these kinds of problems. Prison Dilemma Game is a well-known game that is proposed in game theory. This paper tries to describe how using neural network, the iterated prisoner’s dilemma game can be modeled and analyzed. To do this a Grossberg Counter-Propagation Neural Network (GCP-NN) has been designed to play this game. Results show the capability of this method in complete modeling game. The results present the efficiency of the new method in comparison with the two conventional methods: Tit For Tat (TFT) strategy and Perceptron modeled game. Manuscript profile

Open Access Article

8 - Designing a Self-Tuning Frequency Controller Based on ANNs for an Isolated Microgrid
F. Habibi H. Bevrani J. Moshtag

20.1001.1.16823745.1391.10.3.3.1

Increasing electrical energy demand, as well as fossil fuel shortages and environmental concerns have caused to use uncommon sources such as distributed generations (DGs) and renewable energy sources (RESs) into modern power systems. A microgrid (MG) system consists of More

Increasing electrical energy demand, as well as fossil fuel shortages and environmental concerns have caused to use uncommon sources such as distributed generations (DGs) and renewable energy sources (RESs) into modern power systems. A microgrid (MG) system consists of several DGs and RESs which is responsible to provide both electrical and heat powers for local loads. Due to the MGs nonlinearity/complexity which is imposed to the conventional power systems, classical and nonflexible control structures may not represent desirable performance over a wide range of operating conditions. Therefore, more flexible/intelligent control methods are needed most of the past. Hence, in this paper addresses to design an online/self-tuning PI-controller based on artificial neural networks (ANNs) for optimal regulating the MG systems frequency. Manuscript profile

Open Access Article

9 - Electrical Islanding Detection in Electrical Distribution Networks with Distributed Generation Using Discrete Wavelet Transform and Artificial Neural Network
M. Heidari Orejloo S. Gh. Seifossadat M. Razaz

20.1001.1.16823745.1392.11.1.3.4

In this paper a new algorithm is provided for detecting of electrical islands, based on analysis of transient signals using discrete wavelet transform (DWT) and artificial neural network (ANN). The neural network is taught for Classification of events to the "islands" o More

In this paper a new algorithm is provided for detecting of electrical islands, based on analysis of transient signals using discrete wavelet transform (DWT) and artificial neural network (ANN). The neural network is taught for Classification of events to the "islands" or "non-islands". Needed features for classification are extracted by DWT of DG transient voltage signal. DIgSILENT, MATLAB and WEKA softwares are used for simulation. Proposed method is tested on a CIGRE medium voltage distribution system with two different types of DGs. The final method is chosen from among 162 relay projects with respect to different criteria, including accuracy, speed, simplicity and cost efficiency is the best. With The analysis done in the best relay selection for DGs, the voltage signal, the mother wavelet db4 and seventh level wavelet transform are used. Simulation results show that this method in compared with existing methods, can detect the electrical islands, with a shorter time and higher accuracy. Manuscript profile

Open Access Article

10 - Simulation of Electrical Fault in Stator Winding of Permanent Magnet Synchronous Motor and Discriminating It from Other Possible Electrical Faults Using Probabilistic Neural Network
M. Taghipour-gorjikolaie S. M. Razavi M. A. Shamsi-Nejad

20.1001.1.16823745.1394.13.1.1.6

One of the most common electrical faults in Permanent Magnet Synchronous Motor (PMSM) is inter-turn fault in stator winding. At the incipient steps it seems not dangerous and so light, but spreading this fault can leads to irreparable Consequences. In this paper, the in More

One of the most common electrical faults in Permanent Magnet Synchronous Motor (PMSM) is inter-turn fault in stator winding. At the incipient steps it seems not dangerous and so light, but spreading this fault can leads to irreparable Consequences. In this paper, the intelligent system is presented to protect PMSMs from this kind fault. At the first, intelligent protection system determine the condition of the motor (which can be: Normal, Phase-phase short circuit, Open circuit and Inter-turn fault conditions). If the system determines the faults then send an alarm to operator and also if the fault is inter-turn, it can determine the damaged phase. Obtaining results show that Probabilistic Neural Network can be the most reliable and robust protection system for PMSMs against internal faults, especially inter-turn faults. Manuscript profile

Open Access Article

11 - Application of Wide-Area Synchrophasor Measurement System to Alleviate Blackouts by Rotor Angle Instability
S. Kiarostami S. Kiarostami

20.1001.1.16823745.1395.14.2.7.1

In this paper, a Wide-Area protection system to deal with rotor angle instabilities is proposed. Firstly, a system blackout model is developed and secondly the extreme contingencies that lead to large blackouts are extracted. Initiating events that ultimately lead to ro More

In this paper, a Wide-Area protection system to deal with rotor angle instabilities is proposed. Firstly, a system blackout model is developed and secondly the extreme contingencies that lead to large blackouts are extracted. Initiating events that ultimately lead to rotor angle instabilities are determined by artificial neural network (ANN). Coherent generators are detected by an algorithm using the data presented by phasor measurement units (PMUs). Based on identification of coherent generators, the power system is split into stable islands by disconnecting the weak interconnecting lines and load shedding. The performance of the proposed strategy is verified by simulations on the IEEE 39-bus sample power system. Manuscript profile

Open Access Article

12 - An Efficient Method for Modulation Recognition of MPSK Signals in Fading Channels
S. Hakimi

20.1001.1.16823745.1396.15.2.3.4

Automatic modulation recognition of digital signals is an essential for intelligent communication systems. Most automatic classifications of digital signal types deal with recognizing signals formats in presence of additive white Gaussian noise (AWGN) in channels. Howev More

Automatic modulation recognition of digital signals is an essential for intelligent communication systems. Most automatic classifications of digital signal types deal with recognizing signals formats in presence of additive white Gaussian noise (AWGN) in channels. However, real world communication environments, such as wireless communication channels, suffer from fading effects. There are few methods proposed to perform in fading channels. This paper presents a high efficient method for identification of M-array phase shift keying (MPSK) digital signal type. The proposed method is heuristic hybrid, formed by a multilayer perceptron (MLP) neural network as the classifier and the bees algorithm (BA) as the optimizer. An equalizer is also used to reduce channel effects. A suitable combination of higher order statistics, up to eighth, is considered as prominent characteristics of signals. Simulation results validate the high efficiency of the proposed technique in recognizing the types of digital signals even at low SNRs. Manuscript profile

Open Access Article

13 - Human Action Recognition in Still Image of Human Pose using Multi-Stream neural Network
Roghayeh Yousefi K. Faez

20.1001.1.16823745.1399.18.3.8.2

Today, human action recognition in still images has become one of the active topics in computer vision and pattern recognition. The focus is on identifying human action or behavior in a single static image. Unlike the traditional methods that use videos or a sequence of More

Today, human action recognition in still images has become one of the active topics in computer vision and pattern recognition. The focus is on identifying human action or behavior in a single static image. Unlike the traditional methods that use videos or a sequence of images for human action recognition, still images do not involve temporal information. Therefore, still image-based action recognition is more challenging compared to video-based recognition. Given the importance of motion information in action recognition, the Im2flow method has been used to estimate motion information from a static image. To do this, three deep neural networks are combined together, called a three-stream neural network. The proposed structure of this paper, namely the three-stream network, stemmed from the combination of three deep neural networks. The first, second and third networks are trained based on the raw color image, the optical flow predicted by the image, and the human pose obtained in the image, respectively. In other words, in this study, in addition to the predicted spatial and temporal information, the information on human pose is also used for human action recognition due to its importance in recognition performance. Results revealed that the introduced three-stream neural network can improve the accuracy of human action recognition. The accuracy of the proposed method on Willow7 action, Pascal voc2012, and Stanford10 data sets were 91.8%, 91.02%, and 96.97%, respectively, which indicates the promising performance of the introduced method compared to state-of-the-art performance. Manuscript profile

Open Access Article

14 - An Adaptive Multi-Objective Clustering Algorithm based on Auction_Prediction for Mobile Target Tracking in Wireless Sensor Network
Roghieh Alinezhad Sepideh Adabi arash Sharifi

20.1001.1.16823745.1399.18.4.14.0

One of the applications of sensor networks is to track moving target. In designing the algorithm for target tracking two issues are of importance: reduction of energy consumption and improvement of the tracking quality. One of the solutions for reduction of energy consu More

One of the applications of sensor networks is to track moving target. In designing the algorithm for target tracking two issues are of importance: reduction of energy consumption and improvement of the tracking quality. One of the solutions for reduction of energy consumption is to form a tracking cluster. Two major challenges in formation of the tracking cluster are when and how it should be formed. To decrease the number of messages which are exchanged to form the tracking cluster an auction mechanism is adopted. The sensor’s bid in an auction is dynamically and independently determined with the aim of establishing an appropriate tradeoff between network lifetime and the accuracy of tracking. Furthermore, since the tracking cluster should be formed and activated before the target arrives to the concerned region (especially in high speed of target), avoidance from delay in formation of the tracking cluster is another challenge. Not addressing the mentioned challenge results in increased target missing rate and consequently energy loss. To overcome this challenge, it is proposed to predict the target’s position in the next two steps by using neural network and then, simultaneously form the tracking clusters in the next one and two steps. The results obtained from simulation indicate that the proposed algorithm outperforms AASA (Auction-based Adaptive Sensor Activation). Manuscript profile

Open Access Article

15 - Improving the Architecture of Convolutional Neural Network for Classification of Images Corrupted by Impulse Noise
Mohammad Momeny M. Agha Sarram A. M. Latif R. Sheikhpour

20.1001.1.16823745.1398.17.4.10.9

Impulse noise is one the common noises which reduces the performance of convolutional neural networks (CNNs) in image classification. Preprocessing for removal of impulse noise is a costly process which may have a destructive effect on the training and validation of the More

Impulse noise is one the common noises which reduces the performance of convolutional neural networks (CNNs) in image classification. Preprocessing for removal of impulse noise is a costly process which may have a destructive effect on the training and validation of the convolutional neural networks due to insufficient improvement of noisy images. In this paper, a convolutional neural network is proposed which is robust to impulse noise. Proposed CNN classify images corrupted by impulse noise without any preprocessing for noise removal. A noise detection layer is placed at the beginning of the proposed CNN to prevent the processing of noisy values. The ILSVRC-2012 database is used to train the proposed CNN. Experimental results show that preventing the impact of impulse noise on the training process and classification of CNN can increase the accuracy and speed of the network training. The proposed CNN with error of 0.24 is better than other methods in classification of noisy image corrupted by impulse noise with 10% density. The time complexity of O(1) in the proposed CNN for robustness to noise indicates the superiority of the proposed CNN. Manuscript profile

Open Access Article

16 - Comprehensive Optimal Management System of Distributed Resources Using Dynamic Neural Network in Modeling of Electricity Consumption Uncertainty for Grid-Connected Microgrids
Mohammad Veysi محمدرضا سلطانپور jafar Khalilpour hadi niaei

20.1001.1.16823745.1400.19.1.4.1

In this paper, to enhance the optimal planning for power management of micrigrids, a strategy is proposed using power sharing through coordination between microgrids and the neighborhood system, which has no additional costs for generating units. The uncertainty values More

In this paper, to enhance the optimal planning for power management of micrigrids, a strategy is proposed using power sharing through coordination between microgrids and the neighborhood system, which has no additional costs for generating units. The uncertainty values of electrical consumers are modeled by dynamic neural network, considering the implementation process and high accuracy of forecasting. In another view, to supply the electrical energy of microgrid, diesel generator, renewable energies such as solar energy and wind energy and so, battery energy storage are used, in addition to the upstream grid connection. As well as, using of the reliability factors, along with a detailed assessment of current costs will improve the performance of microgrid. Hence, the loss of power supply probability (LPSP) and loss of load expectations (LOLE) are expressed as factors for assessing the accuracy of current costs. The proposed model is implemented in GAMS and MATLAB environment and the simulation results clearly demonstrate the desired performance of the proposed algorithm, and leads to gaining revenue for the under-study system. Manuscript profile

Open Access Article

17 - Convolutional Neural Networks for Sentiment Analysis in Persian Social Media
M. Rohanian M. Salehi A. Darzi وحید رنجبر

20.1001.1.16823745.1399.18.1.16.6

With the social media engagement on the rise, the resulting data can be used as a rich resource for analyzing and understanding different phenomena around us. A sentiment analysis system employs these data to find the attitude of social media users towards certain entit More

With the social media engagement on the rise, the resulting data can be used as a rich resource for analyzing and understanding different phenomena around us. A sentiment analysis system employs these data to find the attitude of social media users towards certain entities in a given document. In this paper we propose a sentiment analysis method for Persian text using Convolutional Neural Network (CNN), a feedforward Artificial Neural Network, that categorize sentences into two and five classes (considering their intensity) by applying a layer of convolution over input data through different filters. We evaluated the method on three different datasets of Persian social media texts using Area under Curve metric. The final results show the advantage of using CNN over earlier attempts at developing traditional machine learning methods for Persian texts sentiment classification especially for short texts. Manuscript profile

Open Access Article

18 - Performance Improvement of Polynomial Neural Network Classifier using Whale Optimization Algorithm
Mahsa Memari A. Harifi a. Khalili

20.1001.1.16823745.1401.20.2.1.7

Polynomial neural network (PNN) is a supervised learning algorithm which is one of the most popular models used in real applications. The architectural complexity of polynomial neural network in terms of both number of partial descriptions (PDs) and number of layers, le More

Polynomial neural network (PNN) is a supervised learning algorithm which is one of the most popular models used in real applications. The architectural complexity of polynomial neural network in terms of both number of partial descriptions (PDs) and number of layers, leads to more computation time and more storage space requirement. In general, it can be said that the architecture of the polynomial neural networks is very complex and it requires large memory and computation time. In this research, a novel approach has been proposed to improve the classification performance of a polynomial neural network using the Whale Optimization Algorithm (PNN-WOA). In this approach, the PDs are generated at the first layer based on the combination of two features. The second layer nodes consists of PDs generated in the first layer, input variables and bias. Finally, the polynomial neural network output is obtained by sum of weighted values of the second layer outputs. Using the Whale Optimization Algorithm (WOA), the best vector of weighting coefficients will be obtained in such a way that the PNN network reach to the highest classification accuracy. Eleven different dataset from UCI database has been used as input data of proposed PNN-WOA and the results has been presented. The proposed method outperforms state-of-the-art approaches such as PNN-RCGA, PNN-MOPPSO, RCPNN-PSO and S-TWSVM in most cases. For datasets, an improvement of accuracy between 0.18% and 10.33% can be seen. Also, the results of the Friedman test indicate the statistical superiority of the proposed PNN-WOA model compared to other methods with p value of 0.039. Manuscript profile

Open Access Article

19 - Numeric Polarity Detection based on Employing Recursive Deep Neural Networks and Supervised Learning on Persian Reviews of E-Commerce Users in Opinion Mining Domain
Sepideh Jamshidinejad Fatemeh Ahmadi-Abkenari Peiman Bayat

20.1001.1.16823745.1400.19.3.15.6

Opinion mining as a sub domain of data mining is highly dependent on natural language processing filed. Due to the emerging role of e-commerce, opinion mining becomes one of the interesting fields of study in information retrieval scope. This domain focuses on various s More

Opinion mining as a sub domain of data mining is highly dependent on natural language processing filed. Due to the emerging role of e-commerce, opinion mining becomes one of the interesting fields of study in information retrieval scope. This domain focuses on various sub areas such as polarity detection, aspect elicitation and spam opinion detection. Although there is an internal dependency among these sub sets, but designing a thorough framework including all of the mentioned areas is a highly demanding and challenging task. Most of the literatures in this area have been conducted on English language and focused on one orbit with a binary outcome for polarity detection. Although the employment of supervised learning approaches is among the common utilizations in this area, but the application of deep neural networks has been concentrated with various objectives in recent years so far. Since the absence of a trustworthy and a complete framework with special focuses on each impacting sub domains is highly observed in opinion mining, hence this paper concentrates on this matter. So, through the usage of opinion mining and natural language processing approaches on Persian language, the deep neural network-based framework called RSAD that was previously suggested and developed by the authors of this paper is optimized here to include the binary and numeric polarity detection output of sentences on aspect level. Our evaluation on RSAD performance in comparison with other approaches proves its robustness. Manuscript profile

Open Access Article

20 - An Intelligent Vision System for Automatic Forest Fire Surveillance
Mohammad Sadegh Kayhanpanah Behrooz Koohestani

20.1001.1.16823745.1401.20.1.3.7

Fighting forest fires to avoid their potential dangers as well as protect natural resources is a challenge for researchers. The goal of this research is to identify the features of fire and smoke from the unmanned aerial vehicle (UAV) visual images for classification, o More

Fighting forest fires to avoid their potential dangers as well as protect natural resources is a challenge for researchers. The goal of this research is to identify the features of fire and smoke from the unmanned aerial vehicle (UAV) visual images for classification, object detection, and image segmentation. Because forests are highly complex and nonstructured environments, the use of the vision system is still having problems such as the analogues of flame characteristics to sunlight, plants, and animals, or the smoke blocking the images of the fire, which causes false alarms. The proposed method in this research is the use of convolutional neural networks (CNNs) as a deep learning method that can automatically extract or generate features in different layers. First, we collect data and increase them according to data augmentation methods, and then, the use of a 12-layer network for classification as well as transfer learning method for segmentation of images is proposed. The results show that the data augmentation method used due to resizing and processing the input images to the network to prevent the drastic reduction of the features in the original images and also the CNNs used can extract the fire and smoke features in the images well and finally detect and localize them. Manuscript profile

Open Access Article

21 - A POI Recommendation Model According to the Behavior Pattern of Users Based on Friends List Using Deep Learning
sadaf safavi mehrdad jalali

20.1001.1.16823745.1400.19.2.16.5

The rapid growth of Location-based Social Networks (LBSNs) is a great opportunity to provide personalized recommendation services. An important task to recommend an accurate Point-of-Interests (POIs) to users, given the challenges of rich contexts and data sparsity, is More

The rapid growth of Location-based Social Networks (LBSNs) is a great opportunity to provide personalized recommendation services. An important task to recommend an accurate Point-of-Interests (POIs) to users, given the challenges of rich contexts and data sparsity, is to investigate numerous significant traits of users and POIs. In this work, a novel method is presented for POI recommendation to develop the accurate sequence of top-k POIs to users, which is a combination of convolutional neural network, clustering and friendship. To discover the likeness, we use the mean-shift clustering method and only consider the influence of the most similarities in pattern’s friendship, which has the greatest psychological and behavioral impact rather than all user’s friendship. The new framework of a convolutional neural network with 10 layers can predict the next suitable venues and then select the accurate places based on the shortest distance from the similar friend behavior pattern. This approach is appraised on two LBSN datasets, and the experimental results represent that our strategy has significant improvements over the state-of-the-art techniques for POI recommendation. Manuscript profile

Open Access Article

22 - A Step towards All-Optical Deep Neural Networks: Utilizing Nonlinear Optical Element
Aida Ebrahimi Dehghan Pour S. K.

20.1001.1.16823745.1401.20.3.2.0

In recent years, optical neural networks have received a lot of attention due to their high speed and low power consumption. However, these networks still have many limitations. One of these limitations is implementing their nonlinear layer. In this paper, the implement More

In recent years, optical neural networks have received a lot of attention due to their high speed and low power consumption. However, these networks still have many limitations. One of these limitations is implementing their nonlinear layer. In this paper, the implementation of nonlinear unit for an optical convolutional neural network is investigated, so that using this nonlinear unit, we can realize an all-optical convolutional neural network with the same accuracy as the electrical networks, while providing higher speed and lower power consumption. In this regard, first of all, different methods of implementing optical nonlinear unit are reviewed. Then, the impact of utilizing saturable absorber, as the nonlinear unit in different layers of CNN, on the network’s accuracy is investigated, and finally, a new and simple method is proposed to preserve the accuracy of the optical neural networks utilizing saturable absorber as the nonlinear activating function. Manuscript profile

Open Access Article

23 - Stock Price Movement Prediction Using Directed Graph Attention Network
Alireza Jafari Saman Haratizadeh

20.1001.1.16823745.1402.21.1.13.4

Prediction of the future behavior of the stock market has always attracted researchers' attention as an important challenge in the field of machine learning. In recent years deep learning methods have been successfully applied in this domain to improve prediction perfor More

Prediction of the future behavior of the stock market has always attracted researchers' attention as an important challenge in the field of machine learning. In recent years deep learning methods have been successfully applied in this domain to improve prediction performance. Previous studies have demonstrated that aggregating information from related stocks can improve the performance of prediction. However, the capacity of modeling the stocks relations as directed graphs and the power of sophisticated graph embedding techniques such as Graph Attention Networks have not been exploited so far for prediction in this domain. In this work, we introduce a framework called DeepNet that creates a directed graph representing how useful the data from each stock can be for improving the prediction accuracy of any other stocks. DeepNet then applies Graph Attention Network to extract a useful representation for each node by aggregating information from its neighbors, while the optimal amount of each neighbor's contribution is learned during the training phase. We have developed a novel Graph Attention Network model called DGAT that is able to define unequal contribution values for each pair of adjacent nodes in a directed graph. Our evaluation experiments on the Tehran Stock Exchange data show that the introduced prediction model outperforms the state-of-the-art baseline algorithms in terms of accuracy and MCC measures. Manuscript profile

Open Access Article

24 - Data-Driven Sliding Mode Control Based on Projection Recurrent Neural Network for HIV Infection: A Singular Value Approach
Ashkan Zarghami mehdi Siahi Fereidoun Nowshiravan Rahatabad

In the present study, drug treatment of HIV infection is investigated using a Data-Driven Sliding Mode Control (DDSMC) combined with a Projection Recurrent Neural Network (PRNN). The major objective is to establish the control law that eliminates the need for HIV infect More

In the present study, drug treatment of HIV infection is investigated using a Data-Driven Sliding Mode Control (DDSMC) combined with a Projection Recurrent Neural Network (PRNN). The major objective is to establish the control law that eliminates the need for HIV infection mathematical formulae and ensures that the physical limits of the actuator are reached. This is accomplished by creating the concepts of model-free adaptive control, in which the relation between input and output is described using local dynamic linearized models based on quasi-partial derivatives. To determine the DDSMC law, a performance index is first defined based on the fulfillment of a discrete-time exponential reaching condition. By turning this index into a quadratic programming problem, the dynamics of the PRNN are extracted based on projection theory. The closed-loop system is explicitly determined using the optimizer output equation and the closed-loop stability analysis is evaluated using the singular value approach. The simulation results reveal that the proposed algorithm has robust performance in conducting the state variables of HIV infection to the healthy equilibrium point in the face of model uncertainty and external disturbances when compared to one of the newest control techniques. Manuscript profile

Open Access Article

25 - Identification of Cancer-Causing Genes in Gene Network Using Feedforward Neural Network Architecture
مصطفی اخوان صفار abbas ali rezaee

Identifying the genes that initiate cancer or the cause of cancer is one of the important research topics in the field of oncology and bioinformatics. After the mutation occurs in the cancer-causing genes, they transfer it to other genes through protein-protein interact More

Identifying the genes that initiate cancer or the cause of cancer is one of the important research topics in the field of oncology and bioinformatics. After the mutation occurs in the cancer-causing genes, they transfer it to other genes through protein-protein interactions, and in this way, they cause cell dysfunction and the occurrence of disease and cancer. So far, various methods have been proposed to predict and classify cancer-causing genes. These methods mostly rely on genomic and transcriptomic data. Therefore, they have a low harmonic mean in the results. Research in this field continues to improve the accuracy of the results. Therefore, network-based methods and bioinformatics have come to the aid of this field. In this study, we proposed an approach that does not rely on mutation data and uses network methods for feature extraction and feedforward three-layer neural network for gene classification. For this purpose, the breast cancer transcriptional regulatory network was first constructed. Then, the different features of each gene were extracted as vectors. Finally, the obtained vectors were given to a feedforward neural network for classification. The obtained results show that the use of methods based on multilayer neural networks can improve the accuracy and harmonic mean and improve the performance compared to other computational methods. Manuscript profile

Open Access Article

26 - Modeling of Solar Power Plant Using a Neural Network Based on the Equivalent of a Single Diode
Ali Reza reisi Rohollah Abdollahi

<p style="padding-right: 30px; text-align: justify;"><!-- [if gte mso 9]><xml> <w:WordDocument> <w:View>Norma More

<p style="padding-right: 30px; text-align: justify;"></p> <p class="Abstract" style="padding-right: 30px; text-align: justify;"><span style="mso-bidi-font-weight: normal;">Various methods have been proposed for modeling solar panels, but modeling solar power plants using them is associated with challenges. In equivalent circuit-based methods, the modeling depends on factory data that changes over time. Modeling of voltage-current characteristic using intelligent methods such as neural network was less considered due to the low accuracy of modeling. In this article, a method independent of the manufacturer's data for modeling the solar power plant is presented, so that it is possible to accurately model the solar power plants that have been installed for several years. The proposed method consists of two steps, in the first step, open circuit voltage, maximum power point and short circuit current are modeled according to atmospheric conditions using neural network. In the second step, the unknown parameters of the equivalent circuit are determined by circuit analysis relations and using neural network outputs. Finally, to evaluate the proposed method, a 3-kW solar power plant was modeled, and the results show the appropriate accuracy of the proposed method for modeling the solar power plant.</span></p> <p style="padding-right: 90px; text-align: justify;"> </p> Manuscript profile

Open Access Article

27 - Nonlinear Fractional Intelligent Controller for Photovoltaic Inverters
Hadi Delavari Sara Arjmandpour

<p style="text-align: justify; padding-right: 30px;"><!-- [if gte mso 9]><xml> <w:WordDocument> <w:View>Norma More

<p style="text-align: justify; padding-right: 30px;"><span style="font-size: 10.0pt; font-family: 'Times New Roman',serif; mso-fareast-font-family: 'Times New Roman'; mso-ansi-language: EN-US; mso-fareast-language: EN-US; mso-bidi-language: AR-SA;">At present, with the significant growth of energy consumption, increase of greenhouse gases and environmental pollutants, more attention is directed toward renewable energies. Renewable energies include geothermal, wind, photovoltaic energy and etc. Among the advantages of photovoltaic energy, its wide range and easy access, helping to preserve the environment, compatibility with distributed power networks, low noise, quick installation and lower cost compared to other energies can be noted. Important challenges facing photovoltaic systems are changing climatic conditions and parameters variation that affect the performance of the system. In this paper, to track the maximum power point in a photovoltaic system, a fuzzy fractional order sliding mode controller based on disturbance observer and uncertainty estimator using neural network is designed. The sliding mode control is used to reduce chattering, neural network to estimate the system uncertainties, fuzzy system to estimate the coefficient of the signum function in the control law and disturbance observer to approximate the disturbances in the system. Also, the stability of the system has been proven using the Lyapunov method. The simulation results of the photovoltaic system confirm the effectiveness of the proposed method and shows satisfactory performance.</span></p> Manuscript profile

Current Issue

Published Issues

Menu

Browse

List of Articles neural network