نشریه مهندسی برق و مهندسی کامپیوتر ایران
سال بیست و یکم شماره 3 (پیاپی 81، پاییز 1402)

Special conditions of wireless sensor networks, such as energy limitation, make it essential to accelerate the convergence of algorithms in this field, especially in the distributed compressive sensing (DCS) scenarios, which have a complex reconstruction phase. This paper presents a DCS reconstruction algorithm that provides a higher convergence rate. The proposed algorithm is a distributed primal-dual algorithm in a bidirectional incremental cooperation mode where the parameters change with time. The parameters are changed systematically in the convex optimization problems in which the constraint and cooperation functions are strongly convex. The proposed method is supported by simulations, which show the higher performance of the proposed algorithm in terms of convergence rate, even in stricter conditions such as the small number of measurements or the lower degree of sparsity.

The purpose of embedding social networks, which has recently attracted a lot of attention, is to learn to display in small dimensions for each node in the network while maintaining the structure and characteristics of the network. In this paper, we propose the effect of identifying communities in different situations such as community detection during or before the process of random walking and also the effect of semantic textual information of each node on network embedding. Then two main frameworks have been proposed with community and context aware network embedding and community and semantic feature-oriented network embedding. In this paper, in community and context aware network embedding, the detection of communities before the random walk process, is performed through using the EdMot non-overlapping method and EgoNetSplitter overlapping method. However, in community and semantic feature-oriented network embedding, the recognition of communities during a random walk event is conducted using a Biterm topic model. In all the proposed methods, text analysis is examined and finally, the final display is performed using the Skip-Gram model in the network. Experiments have shown that the methods proposed in this paper work better than the superior network embedding methods such as Deepwalk, CARE, CONE, and COANE and have reached an accuracy of nearly 0.9 and better than other methods in terms of edge prediction criteria in the network.

Autism Spectrum Disorder is a type of disorder in which, the patients suffer from a developmental disorder that manifests itself by symptoms such as inability to social communication. Thus, the most apparent sign of autism is a speech disorder. The first part of this paper reviews research studies conducted to automatically diagnose autism based on speech processing methods. According to our review, the main speech processing approaches for diagnosing autism can be divided into two groups. The first group detects individuals with autism by processing their answers or feelings in response to questions or stories. The second group distinguishes people with autism from healthy people because of the accuracy of recognizing their spoken utterances based on automatic speech recognition systems. Despite much research being conducted outside Iran, few studies have been conducted in Iran. The most important reason for this is the lack of rich data that meet the needs of autism diagnosis based on the speech processing of suspected people. In the second part of the paper, we discuss the process of designing, collecting, and evaluating a speaker-independent dataset for autism diagnosis in Iranian children as the first step in the localization of the mentioned field.

Increasing the sharpness of the image, in many cases, refers to strengthening its high frequency components and increasing the sharpness at the edges. In the existing models of increasing clarity, it is assumed that the sensitivity of the human visual system is the same in the whole scene, and the effects of visual attention caused by visual salience are not included in these models. Various studies have shown that visual sensitivity is higher in places that attract more attention. Therefore, increasing image clarity based on visual attention can cause greater perceived clarity in the image. In this article, a model for increasing image sharpness is proposed, which uses the relationship between the map of high frequency image components and visual salience to determine the optimal value of image sharpness. By using a non-linear function, the proposed model expresses the optimal sharpness value for an image according to its visual prominence. Determining the parameters of the nonlinear function in the form of a modeled optimization problem, the solution of which leads to finding the optimal sharpness value automatically. The results show that the proposed method has a more effective performance than the other compared methods if the appropriate values of the control parameters are selected.

Nowadays, quantum computing has played a significant role in increasing the speed of algorithms. Due to the limitations in the manufacturing technologies of quantum computers, the design of a large-scale quantum computer faces many challenges. One solution to overcome these challenges is the design of distributed quantum systems. In these systems, quantum computers are connected to each other through the teleportation protocol to transfer quantum information. Since quantum teleportation requires quantum resources, it is necessary to reduce the number of that. The purpose of this paper is to present a distributed quantum system considering the two goals of balanced distribution of qubits and minimizing the number of teleportation protocols in two levels. In the first level, by presenting a dynamic programming algorithm, an attempt has been made to distribute qubits in a balanced manner and reduce the number of connections between subsystems. According to the output partitioning obtained from the first level, in the second level and in the stage of implementation of global gates, when one of the qubits of this gate is teleported from the home to the desired destination, this qubit may be able to be used by a number of global gates, observing the precedence restrictions and as a result it reduces the number of teleportations. The obtained results show the better performance of the proposed algorithm.

By using the line detection system, the relative position of the self-driving cars compared to other cars, the possibility of leaving the lane or an accident can be checked. In this paper, a fast and lightweight line detection approach for images taken from a camera installed in the windshield of cars is presented. Most of the existing methods consider the problem of line detection in the form of classification at the pixel level. These methods despite having high accuracy, suffer from two weaknesses of having the high computational cost and not paying attention to the general lines content information of the image (as a result, they cannot detect if there is an obstacle). The proposed method checks the presence of lines in each row by using the row-based selection method. Also, the use of Mobile-net architecture has led to good results with fewer learning parameters. The use of three different functions as cost functions, with different objectives, has resulted in obtaining excellent results and considering general content information along with local information. Experiments conducted on the TuSimple video image collection show the suitable performance of the proposed approach both in terms of efficiency and especially in terms of speed.

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.

Physically unclonable functions (PUFs) provide hardware to generate a unique challenge-response pattern for authentication and encryption purposes. An essential feature of these circuits is their unpredictability, meaning that an adversary cannot sufficiently predict future responses from previous observations. However, machine learning algorithms have been demonstrated to be a severe threat to PUFs since they are capable of accurately modeling their behavior. In this work, we analyze PUF security threats and propose a PUF-based authentication mechanism called SQ-PUF, which can provide good resistance to machine learning attacks. In order to make it harder to simulate or predict, we obfuscated the correlation between challenge-response pairs. Experimental results show that, unlike existing PUFs, even with a large data set, the SQ-PUF model cannot be successfully attacked with a maximum prediction accuracy of 53%, indicating that this model is unpredictable. In addition, the uniformity in this model remains almost the same as the ideal value in A-PUF.