Presenting a Personalized Web Recommender System Based on Sequential Clustering and Deep Auto-Encoder

per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-08231116articlePresenting a Personalized Web Recommender System Based on Sequential Clustering and Deep Auto-EncoderM. Moeinist_m_moini@azad.ac.ir1Ali BroumandniaBroumandnia@gmail.com2Mona Moradimona.moradi@iau.ac.ir3Dept. of Software Eng., South Tehran Branch, Islamic Azad University, Tehran IranDept. of Software Eng., South Tehran Branch, Islamic Azad University, Tehran IranDept. of Software Eng., Central Tehran Branch, Islamic Azad University, Tehran Iran South Tehran BranchThe amount of information published on the web has made the use of recommender systems inevitable. Web recommender systems provide users with accurate and relevant recommendations based on their interests and tastes. These recommendations can help users quickly access the information they need and reduce search time. In this paper, a hybrid recommender system based on the combination of fuzzy sequential clustering and deep Auto-encoder network based on user profile information and ranking of websites by users is presented. In this recommender system, users are first sequentially clustered according to the similarity of their opinions. Then the new ranking for users is predicted according to the fuzzy membership function. Finally, the information in the user profile and the new rating of users to each website is used as the input of the provided deep Auto-encoder network in order to predict the ranking of websites by users. Finally, after finding similar users, It provides recommendations to visit and personalize the web page of new users based on the favorite websites of similar users. The proposed method has improved compared to the following classification methods due to the layers of deep learning and completion of the learning process in the middle layer: In terms of statistical accuracy, about 42%, and the ratio of successful recommendations to useful recommendations is about 4%, and the accuracy of recognizing similar users is about 20%.http://ijece.org/Article/47034Recommender system user profile auto-encoder networks collaborative filter.per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-082311728articleRelation Detection of Persian Questions by Combining Direct and Indirect MethodsAbbas Shahini Shamsabadiabbasshahini60@gmail.com1Reza Ramezaniramezani.cs@gmail.com2Hadi Khosravi farsanikhosravi@eng.sku.ac.ir3Mohammadali nematbakhshnematbakhsh@eng.ui.ac.ir4Department of Software Engineering, Faculty of Computer Engineering, University of Isfahan, IranDepartment of Software Engineering, Faculty of Computer Engineering, University of Isfahan, IranDepartment of Computer Engineering, Shahrekord University, Shahrekord, IranDepartment of Software Engineering, Faculty of Computer Engineering, University of Isfahan, IranIn this study, for the problem of answering Persian questions using linked data, the sub-problem of relation detection for single-relation questions has been investigated in detail. In these questions, the answer is extracted from a triple in the form of <subject, predicate, object>. This process has two main steps: entity linking and relation detection. In the first step, the entity identified in the question is mapped to a subject or object of a triple, and in the second step, a predicate is selected for the semantic relation in the question. In most existing methods, after entity linking, all relations of that entity in the knowledge base are considered as candidate relations, and finally one of them is selected as the final relation. In these methods, if there is an error in the entity linking step, it is propagated to the relation detection step. In this study, to solve this dependency, the hierarchical structure of relations is used in order to directly extract the relation of the question. The accuracy of the proposed method in Persian is 72% for direct relation detection and 90% for selecting the best candidate relation (indirect). The accuracy has increased to 94% by combining direct and indirect methods.  http://ijece.org/Article/47728Persian question-answering relation detection knowledge base natural language processing.per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-082312940articleA Novel Method Based-on Gradient and Deep Neural Network Filters to Generate Texture ImagesM. H. Shakoormhshakoor@gmail.com1Dept. of Comp. Eng., Faculty of Eng., Arak University, Arak, IranProduction of image databases is one of the necessities of machine vision. There are various methods such as rotating, changing the viewing angle, resizing, etc., to increase the image data. The disadvantage of these methods is that the generated images are very similar to the original images and it is not enough to prevent overfitting. Among all types of images, texture images have more challenges. In this research, a new texture is generated using the convolution coefficients of pre-trained deep networks. In this method, new textured images are artificially produced by applying an ascending gradient to the images resulting from convolution filters. The difference between this method and the generative methods is that there is no initial texture image to increase, but here a new class of texture image is generated from the coefficients of the pre-trained deep network. After the new texture is produced, its number is increased by image processing methods. This method is between 3 and 5 times faster than some well-known generator networks. The quality of the images is much better. With this method, a texture database example has been produced, which includes 2400 images in 80 classes, and has been uploaded to the Kaggle site.http://ijece.org/Article/47082Image augmentation data generation gradient ascent convolutional neural network. per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-082314150articleOptimization of Response Time in Software Defined IoT Networks Using Cloud-Fog ComputingElham Hajianelyhaj@gmail.com1Dept. of Comp. Eng. Univeristy of Bojnord, Bojnord, IranThe Internet of Things uses the cloud to process information received from electronic devices. Powerful servers located far from the sensors perform the processing. IoT devices send requests to the cloud and receive results from it. In some IoT applications, response time and latency are important. Therefore, latency should be reduced as much as possible. Sending information to the cloud itself entails latency. Therefore, the use of fog along with the cloud plays a fundamental role in the IoT. The use of fog and cloud computing in the field of IoT is a significant topic for researchers. To facilitate this process, software-defined networks have emerged as a vital component. These networks enable centralized control and management of the network. They also ensure optimal resource utilization and seamless connectivity by dynamically directing data flows to fog or cloud resources based on real-time conditions. Fog computing refers to the deployment of resources near the network sensor. By doing so, fog computing aims to reduce latency and bandwidth usage while improving overall system performance. It does this by utilizing local computing capabilities to process data. This research uses the proposed architecture for IoT networks and modeling different parts of this architecture using queuing theory to reduce response time using fog-cloud computing and obtain network quality of service parameters through mathematical analysis. In the following, the residual energy and latency comparison graphs for health and lighting applications as well as the use and non-use of fog-cloud computing are plotted. The graphs show that the use of fog-cloud computing reduces response time and the lighting application using fog has more residual energy and the health application has less latency. The simulation was performed using NS2 software in the smart home application.http://ijece.org/Article/47481Internet of things software defined networks smart home fog computing cloud. per Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-082315160articleA low-power approximate accelerator based on FPGA chips for artificial intelligence applicationsNadia Sohrabinadiasohrabi@aut.ac.ir1Amir Bavafa Toosiabavafat@yahoo.com2Mehdi Sedighimsedighi@aut.ac.ir3Comp. Eng. Faculty, Amir Kabir University of Technology, Tehran, IranFaculty of Com. Eng., Sadjad University, Mashhad, IranComp. Eng. Faculty, Amir Kabir University of Technology, Tehran, IranOne of the challenges of neural networks is the high calculations. For this reason, many architectures have been proposed for such applications, which provide solutions for their complex calculations. Reconfigurable hardware accelerators such as FPGA are usually used to accelerate neural network; But the main problem of these chips is their relatively high-power consumption. To reduce the power consumption in FPGA, the approximate calculation technique can be used. The main idea of approximate computing is to make  compromise between accuracy and energy consumption by making changes in the circuit or code. In this research, a convolutional neural network has been designed and implemented to recognize handwritten digits in an accurate and approximate manner with the aim of improving the power consumption. This method reduces the power consumption by preventing the transmission of  transfer digit in the low bits of the adder. The results of the comparison of the neural network accurately and approximately show that by approximating the 6 bits of the low weight of the adder, the power consumption is reduced by 43% and no error occurs. Also, by approximating 7 bits of low weight, with 20% error, the power consumption is reduced by 44.11%http://ijece.org/Article/48039Approximate adder convolutional neural network handwritten digit recognition approximate calculationsper Iranian Research Institute for Electrical Engineeringفصلنامه مهندسی برق و مهندسی کامپيوتر ايران16823745168237452025-082316175articleThe Deadline- and Energy-Aware Resource Allocation using a Combination of Multi-Criteria Greedy Approach and Decision Tree in the IoT-Edge-Cloud Environmentshiva Razzaghzadehshiva.razzaghzadeh@gmail.com1Sara Hoseynpoursara.hsnpoor@gmail.com2Dept. of Com. Eng., Ardabil Branch, Islamic Azad University, Ardabil, IranDept. of Com. Eng., Ardabil Branch, Islamic Azad University, Ardabil, IranWith the rapid growth of the Internet of Things (IoT), the volume of data collected from sensors has increased significantly. As a result, there is a growing need to connect IoT devices to cloud servers to meet the demands of data storage, processing, and analysis. Furthermore, the emergence of intermediate technologies, such as fog computing, which performs initial computations on requests at the network edge, has reduced the computational load sent to the cloud. However, task scheduling in cloud resources remains a challenging problem. Resource scheduling, as an NP-Hard problem, involves the optimal and efficient allocation and distribution of resources (such as processors, memory, networks, etc.) to tasks in cloud servers. Therefore, many researchers have attempted to propose heuristic-based algorithms to find near-optimal solutions. In these approaches, the primary goal is to find the appropriate resource for task allocation, while the task’s execution deadline is not always considered. In IoT applications, the data may correspond to critical tasks that require quick responses, which has often been overlooked in previous methods. Therefore, this paper proposes a resource allocation approach using scheduling in the IoT-Fog-Cloud framework, based on a combination of decision trees for task prioritization and a multi-criteria greedy approach. Simulation results show that the proposed method, by prioritizing tasks and balancing multiple objectives using a multi-criteria greedy approach, performs near-optimally in terms of evaluation criteria such as cost and task completion time, and improves upon previous methods.http://ijece.org/Article/47217Internet of things (IoT) resource allocation scheduling decision tree multi-criteria greedy approach.