مطالب

Automating the processing of videos in applications such as surveillance, sport commentary and activity detection, human-machine interaction, and health/disability care is crucial to their correct functioning. In such video processing tasks, recognition of various human actions is a pivotal component for the correct understanding of videos and making decisions upon it. Accurately recognizing human actions is a complex process, demanding high computing capabilities and intelligent algorithms. Several factors, such as object occlusion, camera movement, and background clutter, further challenge the task and its accuracy, essentially leaving deep learning approaches the only viable option for properly detecting human actions in videos. In this study, we propose CoReHAR, a novel Human Action Recognition method that employs both deep Convolutional and Recurrent neural networks on raw video frames. Using the pre-trained ResNet152 CNN, deep features are initially extracted from video frames. The sequential information of the frames is then learned using DB-LSTM RNN. Multiple stacked layers in forward and backward passes of the DB-LSTM provide increased network depth for higher accuracy. A number of techniques are also applied to improve CoReHAR’s processing speed on heterogeneous GPU-enabled systems. The proposed method is evaluated using PyTorch, and is compared to the state-of-the-art methods, showing a considerable efficiency increase, with nearly 95% recognition accuracy measured as an average over all splits of the challenging UCF101 dataset.

Today, due to the large volume of data and the high speed of data production, it is practically impossible to analyze data using traditional methods. Meanwhile, data mining, as one of the most popular topics in the present century, has contributed to the advancement of science and technology in a number of areas. In the recent decade, researchers have made extensive use of data mining to analyze data. One of the most important issues for researchers in this field is to identify common mainstreams in the fields of data mining and to find active research fields in this area for future research. On the other hand, the analysis of social networks in recent years as a suitable tool to study the present and future relationships between the entities of a network structure has attracted the researcher’s scrutiny. In this paper, using the method of co-occurrence analysis of words and analysis of social networks, the scientific structure and map of data mining issues in Iran based on papers indexed during the years 1388 to 1398 in the Civilica database is drawn, and the thematic trend governing research in this area has been reviewed. The results of the analysis show that in the category of data mining, concepts such as clustering, classification, decision tree, and neural network include the largest volume of applications such as data mining in medicine, fraud detection, and customer relationship management have had the greatest use of data mining techniques.

Software testing is an expensive and time-consuming process. These costs can be significantly reduced using automated methods. Recently, many researchers have focused on automating this process using search algorithms. Many different methods have been proposed, all of which using a means of heuristic or meta-heuristic search algorithms. The main problem with these methods is that they are usually stuck in local optima. In this paper, to overcome such a problem, we have combined the firefly algorithm (FA) and asexual reproduction optimization algorithm (ARO). FA is a bio-inspired algorithm that is very efficient at exploitation and local searches; however, it suffers from poor exploration and is prone to local optima problem. On the other hand, ARO can be used for escaping from local optima. For this combination, we have inserted ARO into the steps of FA for increasing the population diversity. We have utilized this combination for automatic test case generation with the aim of covering all finite paths of the control flow graph. To evaluate the performance of the proposed method, we have utilized it for generating test cases for a number of programs. Results have indicated that, while giving similar results in terms of the test coverage, the proposed method is significantly better than the existing state of the art algorithms in terms of the number of fitness evaluations. Compared algorithms are FA, ARO, traditional genetic algorithm (TGA), adaptive genetic algorithm (AGA), adaptive particle swarm optimization (APSO), hybrid genetic tabu search algorithm (HGATS), random search (RS), differential evolution (DE), and hybrid cuckoo search and genetic algorithm (CSGA).

The smart home is an important Internet of Things applications. Due to the smartphones development, expansion of their network, and growing the data transfer rate, security in personal life has become a dramatic challenge. Therefore, it is essential to secure such a system to create a sense of relaxation in the lives of users and homeowners to deal with possible occurrences. The integration of technologies for the automation of home affairs with the Internet of things means that all physical objects can be accessed on cyberspace; therefore, the concerns raised by users about the lack of privacy and security are serious arguments that science and technology should answer. Therefore, addressing security issues is a crucial necessity for the development of the smart homes. Although authentication protocols have been proposed based on smart cards for multi-server architectures, their schemes cannot protect the system against stolen smart cards and dictionary attacks in the login phase and do not satisfy perfect forward secrecy. To overcome these limitations, this paper proposes an anonymous, secure protocol in connected smart home environments, using solely lightweight operations. The proposed protocol in this paper provides efficient authentication, key agreement, and enables the anonymity of devices and unlinkability. It is demonstrated that the computation complexity of the protocol is low as compared to the existing schemes, while security has been significantly improved. This protocol ensures that even if the stakeholder’s device or the IoT device is attacked, they are robust against them.