۱.
Network communication shows a variety of issues with the fast expansion of computer devices, ranging from network administration to traffic engineering. A well-known method for improving these connections is Software-Defined Networking (SDN). The SDN is a networking architecture that separates the control plane from the data plane to ease network administration. The main advantage of the SDN is the central controller. However, it has security flaws like unreachability in Distributed Denial-of-Service attacks (DDoS). Hence, defending SDN against DDoS attacks is critical. We proposed a framework for detecting DDoS attacks and a fault-tolerant method to replace faulty leader controller in distributed multi-controller SDN. We used multi-controllers architecture and leader election algorithm to present a fault-tolerant framework to select a new leader controller, in the case of a leader controller failure. In addition, an early DDoS attack detection algorithm using the entropy of destination IP addresses and the packet window initiation rate is presented. To evaluate our proposed method in various configurations, we simulated exhaustive experiments in Mininet and Floodlight. The results show that our approach outperforms similar algorithms in various network configurations and multi-victim attacks.
۲.
COVID-19 is a new variant of SARS-COV-2 which can lead to mild to severe infection in humans. Despite the remarkable efforts to contain the epidemic, the virus spread rapidly around the world and its prevalence continued with different degrees of clinical symptoms in many countries. Although common strategies including prevention, diagnosis, and care are necessary to curb this epidemic, early and accurate diagnosis can play an important role in reducing the speed of the epidemic. In this regard, the use of technologies based on artificial intelligence can be of great help. For this reason, since the outbreak of COVID-19, many researchers have tried to use machine learning techniques as a subset of artificial intelligence for the early diagnosis of COVID-19. Considering the importance and role of using clinical and laboratory data in the diagnosis of people with covid-19, in this paper K-NN, SVM, decision tree, random forest, Naive Bayes, neural network and XGBoost models are the most common machine learning models, and a dataset containing 1354 records consisting of clinical and laboratory data of patients in Imam Hossein Hospital in Tehran has been used to diagnose patients with covid-19. The results of this research indicate that based on the evaluation criteria, XGBoost and K-NN models have the most accuracy among the mentioned models and can be considered suitable predictive models for the diagnosis of COVID-19.
۳.
Review spam is an opinion written to promote or demote a product or brand on websites and other internet services by some users. Since it is not easy for humans to recognize these types of opinions, a model can be provided to detect them. In recent years, much research has been done to detect these types of reviews, and with the expansion of deep neural networks and the efficiency of these networks in various issues, in recent years, multiple types of deep neural networks have been used to identify spam reviews. This paper reviews the proposed deep learning methods for the problem of review spam detection. Challenges, evaluation criteria, and datasets in this area are also examined.
۴.
Community Question Answering has a crucial role in almost all societies nowadays. It is important for the owners of a community to be able to make it better and more reliable. One way to achieve this, is to find the users who have more knowledge, expertise, experience and skill and can well share their knowledge with others (which we call experts and aim to encourage them to be more active in the website). One method to use is to identify expert users, and whenever a new question is asked, we suggest this question to them to check and answer if its in their area of expertise. One way to encourage users to post replies, is to use gameplay techniques such as assigning points and badges to users. But as we will discuss, this method does not always detect expert users well, because some users will try to have small and insignificant but numerous activities that will make them gain a lot of points, however they are not experts. In this study, we examine the methods by which experts in a question-and-answer system can be found, and try to evaluate and compare these methods, use their ideas and positive points, and add our own new ideas to a new way of finding them. We used some ideas such as profile making for users, categorize users’ expertise, A-Priori algorithm and showed that neural networks method results the best for the purpose of expert detection.
۵.
The transportation networks analysis aims to investigate the system's structural characteristics and dynamical evolution to evaluate the transit services. In this regard, the topological characteristics of the Iran railway network have been studied and compared to two well-studied railway networks, China and Spain. Also, the network vulnerability to station failures has been studied based on different attacks. Accordingly, in the first step of this work, the city stations have been extracted from Iran railway information to construct the network. Then, some structural properties, including the degree distribution, betweenness centrality, clustering coefficient, and distance distribution, have been analyzed for three networks. Finally, the network reliability has been evaluated using a random as well as adversarial attack. The structural analysis reveals that the Iran railway network would require some structural optimization to improve the economic benefits. Based on the vulnerability investigation, the network efficiency of the network will be dropped more quickly utilizing the maximum betweenness attack. In addition, as it were, little parts of the network seem to keep their usefulness as the estimate of the giant component is diminished exceptionally strongly when less than 20% of nodes are expelled from the network haphazardly or intentioned.
۶.
The random walk technique, which has a reputation for excellent performance, is one method for complex networks sampling. However, reducing the input data size is still a considerable topic to increase the efficiency and speed of this algorithm. The two approaches discussed in this paper, the no-retracing and the seed node selection algorithms, inspired the development of random walk technique. The Google PageRank method is integrated with these different approaches. Input data size is decreased while critical nodes are preserved. A real database was used for this sampling. Significant sample characteristics were also covered, including average clustering coefficient, sampling effectiveness, degree distribution, and average degree. The no-retracing method, for example, performs better. The efficiency increases even further when the no-retracing technique is combined with the Google PageRank. When choosing between public transportation and aircraft, for example, these algorithms might be used since time is crucial. Additionally, these algorithms are more energy-efficient methods that were looked at.
۷.
Since the outbreak of Covid19 virus to date, various methods have been introduced in order to diagnose the virus infection. One of the most reliable tests is assessing frontal Chest X-Ray(CXR) images. As the virus causes inflammation in the infected patient's lung, it is possible to diagnose whether one is infected or not using his/her CXR image. in contrast to other tests which mostly are based on the virus genome, this test is not time-consuming and it is reliable against new strains of the virus. However, this test requires a specialist to assess the CXR images. As the datasets of Covid19 patient CXR images are increasing in number, it is possible to use machine learning techniques in order to assess CXR images automatically and even online.
In this study, we used deep learning approaches and we fine-tuned the Alexent in order to automatically classify CXR images and label the whether "Covid" or "Normal". The data we used in this study include about 10,000 chest images, half of which are related to CXR images and the other half are related to patients with Covid19 infection. The model proved to be very reliable with 99.26% accuracy in diagnosis and 95% sensitivity and 99.7% specificity.
۸.
Spams are well-known examples of unsolicited text or messages which are sent by unknown individuals and cause issues for smartphone users. The inconvenience imposed on users, the loss of network traffic, the rise in the calculated cost, occupying more physical space on the mobile phone, and abusing and defrauding recipients are but a few of their downsides. Consequently, the automated identification of suspicious and spam messages is undoubtedly vitally important. Additionally, text messages which are smartly composed might be difficult to recognize. However, the present methodologies in this subject are hindered by the absence of adequate Persian datasets. A huge body of research and experiments has revealed that techniques based on deep and combined learning are superior at identifying unpleasant text messages. This work sought to develop an effective strategy for identifying SMS spam through utilizing combining machine learning classification algorithms together with deep learning models. After applying preprocessing on our gathered dataset, the suggested technique applies two convolutional neural network layers, the first of which being an LSTM layer, and the second one which is a fully connected layer to extract the data characteristics, thereby implementing the suggested deep learning approach. As part of the Machine Learning methodologies, the vector support machine makes use of the data and features at hand to determine the ultimate classification. Results indicate that the suggested model is implemented more effectively than the existing techniques, and an accuracy of 97.7% was achieved as a result.
۹.
The majority of wireless sensor network (WSN) security protocols state that a direct connection from an attacker can give them total control of a sensor node. A high level of security is necessary for the acceptance and adoption of sensor networks in a variety of applications. In order to clarify this issue, the current study focuses on identifying abnormalities in nodes and cluster heads as well as developing a method to identify new cluster heads and find anomalies in cluster heads and nodes. We simulated our suggested method using MATLAB tools and the Database of the Intel Research Laboratory. The purpose of the performed simulation is to identify the faulty sensor. Using the IBRL database, sensors that fail over time and their failure model is the form that shows the beats in the form of pulses, we find out that the sensor is broken and is of no value. Of course, this does not mean that the sensor is invasive or intrusive. We have tried by clustering through Euclidean distance that identify disturbing sensors. But in this part of the simulation, we didn't have any data that shows disturbing sensors, it only shows broken sensors. We have placed the sensors randomly in a 50 x 50 space and we want to identify the abnormal node.
۱۰.
In recent years, with the advancement of information technology in education, e-learning quality promotion has received increased attention. Numerous criteria exist for promoting learning quality, such as fitness for purpose, which refers to the extent to which service fits its intended purpose. Multiple purposes are considered in e-learning. One is reducing the knowledge gap between the learner’s perception of educational concepts and what should be understood of training concepts. Identifying and calculating the learner’s knowledge gap is the first step in reducing the knowledge gap. Consequently, this paper presents a new method for calculating the learner’s knowledge gap concerning each concept in the training video content based on the learner’s click behavior. The association between the learner’s knowledge gap and click behavior was determined by categorizing the learner’s click behaviors. Similarly, the Apriori algorithm extracted rules for each behavioral category. The results demonstrated that learning outcome correlated with the learner’s click behavior. Therefore, four behavioral rules regarding the compatibility between the knowledge gap and learner’s click behavior are presented. Experiments were performed by 52 students enrolled in the micro-processing course at Tehran University’s e-Learning Center.
۱۱.
A recommendation system is a system that, based on a limited amount of information provided by users as well as the feedback given to goods, persons, and locations by other users, provides appropriate suggestions to the user. Today, with the large number of physicians and specialists, it seems necessary to have a system for identifying the right specialist and experienced physician for the patient. We present in this study a system for medical recommendations that analyzes physicians and specialists. It uses collaborative filtering and scores provided by other users to suggest physician recommendations according to the area of expertise of the physician. Research conducted and evaluation of results show that this system can successfully recommend a specialist doctor to the user in 90% of cases.
۱۲.
Divorce will have destructive spiritual and material effects, and unfortunately, in this regard recent statistics have shown that solutions provided for its prevention and reduction have not been effective. One of the effective solutions to reduce divorce in society is to review the background of the couple, which can provide valuable experiences to experts, and used by experts and family counselors. In this article, a method has been proposed that uses data mining and deep learning to help family counselors to predict the outcome of marriage as a practical tool. Reviewing the background of thousands of couples will provide a model for the coupe behavior analysis. The primary data of this study was collected from the information of 35,000 couples registered in the National Organization for Civil Registration of Iran during 2018-2019. In the current work, we proposed a method to predict divorce by combining a convolutional neural network (CNN) and long short-term memory (LSTM). In this hybrid method, key features in a dataset are selected using CNN layers, and then predicted using LSTM layers with an accuracy of 99.67 percent. A comparison of the method used in this article and Multilayer Perceptron (MLP) and CNN suggests that it has a higher degree of accuracy.
