نشریه مهندسی برق و مهندسی کامپیوتر ایران
سال بیست و یکم شماره 1 (پیاپی 77، بهار 1402)

In recent years, illegally copying digital images and even manipulating them, without great loss of quality and at a low cost has been made possible. Watermarking has recently been developed as one of the methods to detect that tampering has occurred and even enable some recovery of the original images. However, there are still many issues to resolve in providing an effective watermark that can detect and recover a wide range of manipulations. Furthermore, the accuracy of detecting and the capability of the recovery of the original images by existing methods are still not at an acceptable level. These problems are more critical when certain high-rate manipulation attacks occur. In this paper, a watermarking method will be introduced that not only is able to detect any tampering, but also can successfully recover the original images in high quality, even at high tampering rates. In this method, Singular Value Decomposition (SVD) is used to detect tampering and Optimal Iterative Block Truncation Coding (OIBTC) has also been applied to recover lost data. This paper proposes a powerful way to increase detection sensitivity while increasing watermark resistance for the effective recovery of corrupted images. The results prove the superiority of the proposed method over current methods.92% of tasks are executed successfully in the edge environment.

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.

With the development of telecommunication and communication technologies, especially wireless communications, information cryptography is one of the communication necessities. Today, cryptographic algorithms are used to increase security and prevent DICOM medical images from unauthorized access. It should be noted that changes in DICOM medical images will cause the doctor to misdiagnose the patient's treatment process. In this paper, a type of hybrid cryptographic algorithms is designed. In the proposed algorithm, DNA encryption algorithm is used to encrypt DICOM images and patient biometric information such as fingerprint or iris image is used to make digital signature and validate DICOM medical images. The designed encryption algorithm is resistant to brute force attacks and the entropy of the encrypted DICOM images is above 7.99.

VLSI technology is currently dealing with a serious challenge, as the exponential growth of density in VLSI and CMOS chips has reached its limit. Power dissipation in VLSI chip refers to heat generation, which is a real barrier against traditional CMOS technology. Irreversible logic leads to problems such as energy dissipation, heat generation, information loss and slow computations. We need a new technology for solving these problems. Using reversible logic can help solve this problem. In next generation of optical computers, electrical circuits and wires will be replaced by several optical fibers and these systems will be more efficient because they will be cheaper, lighter, and more compact without interference. Based on optical computations, several optical switches have been proposed for future applications. One of these switches is the Mach-Zehnder switch. Its behavior and the reversible circuits, which can be made with this switch is studied in this article. Finally, we introduce and design three new all-optical reversible gates named NFT, SRK and MPG, which are effective in designing all-optical reversible logical circuits such as flip-flops and other all-optical reversible sequential circuits. We also simulate one all-optical reversible circuits implemented with Mach-Zehnder switch and provide simulation challenges and solutions to overcome these challenges.

Text generation is a field of natural language processing. Text generation enables the system to produce comprehensive, .grammatically correct texts like humans. Applications of text generation include image Captioning, poetry production, production of meteorological reports and environmental reports, production of business reports, automatic text summarization, .With the appearance of deep neural networks, research in the field of text generation has change to use of these networks, but the most important challenge in the field of text generation using deep neural networks is the data is discrete, which has made gradient inability to transmit. Recently, the use of a new approach in the field of deep learning, called generative adversarial networks (GANs) for the generation of image, sound and text has been considered. The purpose of this research is to use this approach to generate Persian sentences. In this paper, three different algorithms of generative adversarial networks were used to generate Persian sentences. to evaluate our proposed methods we use BLEU and self-BLEU because They compare the sentences in terms of quality and variety.

Calculation and applying trust among users has become popular in designing recommender systems in recent years. However, most of the trust-based recommender systems use only one factor for estimating the value of trust. In this paper, a multi-factor approach for estimating trust among users of recommender systems is introduced. In the proposed scheme, first, users of the system are clustered based on their similarities in demographics information and history of ratings. To predict the rating of the active user into a specific item, the value of trust between him and the other users in his cluster is calculated considering the factors i.e. time, location, and context of their rating. To this end, we propose an algorithm based on beta distribution. A novel tree-based measure for computing the semantic similarity between the contexts is utilized. Finally, the rating of the active user is predicted using weighted averaging where trust values are considered as weights. The proposed scheme was performed on three datasets, and the obtained results indicated that it outperforms existing methods in terms of accuracy and other efficiency metrics.

In this paper, a new algorithm is presented to drastically reduce communication overhead in distributed (decentralized) single target tracking in a wireless sensor network. This algorithm is based on a new approach to solving the average consensus problem and the use of distributed particle filters. For the algorithm of this paper, unlike the common algorithms that solve an average consensus problem just to approximate the global likelihood function to calculate the particle importance weights in distributed tracking, a new model for observation is presented based on the Gaussian approximation, which only solves the problem Consensus is applied to the mean on the received observations of the nodes in the network (and not to approximate the global likelihood function). These innovations significantly reduce the exchange of information between network nodes and as a result uses much less energy resources. In different scenarios, the efficiency of the proposed algorithm has been compared with the centralized algorithm and the distributed algorithm based on the graph, and the simulation results show that the communication overhead of the network is greatly reduced in exchange for an acceptable drop in tracking accuracy by using our proposed algorithm.