Vol. 42 · No. 4

Model checking is a software automation verification technique. However, the complex execution process of concurrent software systems and the exhaustive search of state space make the model-checking technique limited by the state-explosion problem in real applications. Due to the uncertain input information (called system parameterization) in concurrent software systems, the state-explosion problem in model checking is exacerbated. To address the problem that reachability graphs of Petri net are difficult to construct and cannot be explored exhaustively due to system parameterization, this paper introduces parameterized variables into the program dependence net (a concurrent program model). Then, it proposes a parameterized reachability graph generation algorithm, including decision algorithms for verifying the properties. We implement LTL- X verification based on parameterized reachability graphs and solve the problem of difficulty constructing reachability graphs caused by uncertain inputs.

Feature selection (FS) averts the consideration of unwanted features which may tend the classification algorithm to classify wrongly. Choosing an optimal feature subset from the given set of features is challenging due to the complex associations present within the features. In non-convex conditions, the gradient-based algorithms suffer due to local optima or saddle points with respect to initial conditions where swarm intelligence algorithms pose a higher chance to converge over the global optima. The Salp Swarm Algorithm (SSA) proposed by Mirjalili et al. is based on the chaining behaviour of sea salps but the algorithm lacks diversity in the exploration stage. Rectifying the exploratory behaviour and testing the algorithm against the FS problem is the motivation behind this work. Three variants of the algorithm are proposed, of which the Vigilant Salp Swarm Algorithm (VSSA) inherits the vigilant mechanism in Grey Wolf Optimizer (GWO), the second variant and the third variant replace a simple crossover operator and shuffle crossover operator instead of the follower's position update mechanism used in the VSSA to form Vanilla Crossover VSSA (VCVSSA) and Shuffle Crossover VSSA (SCVSSA).

In recent years, autonomous driving technology has gradually appeared in our field of vision. It senses the surrounding environment by using radar, laser, ultrasound, GPS, computer vision and other technologies, and then identifies obstacles and various signboards, and plans a suitable path to control the driving of vehicles. However, some problems occur when this technology is applied in foggy environment, such as the low probability of recognizing objects, or the fact that some objects cannot be recognized because the fog's fuzzy degree makes the planned path wrong. In view of this defect, and considering that automatic driving technology needs to respond quickly to objects when driving, this paper extends the prior defogging algorithm of dark channel, and proposes UDP-YOLO network to apply it to automatic driving technology. This paper is mainly divided into two parts: 1. Image processing: firstly, the data set is discriminated whether there is fog or not, then the fogged data set is defogged by defogging algorithm, and finally, the defogged data set is subjected to adaptive brightness enhancement; 2. Target detection: UDP-YOLO network proposed in this paper is used to detect the defogged data set. Through the observation results, it is found that the performance of the model proposed in this paper has been greatly improved while balancing the speed.

This paper proposes an advanced expert collaborative filtering recommendation algorithm. Although ordinary expert system filtering algorithms have improved the recommendation accuracy of collaborative filtering technology to a certain extent, they have not screened the level of expertise of experts, and the credibility of experts varies. Therefore, this paper proposes an expert mining system based on signal fluctuations. The algorithm uses signal processing technology to filter the level of experts. This method introduces a kurtosis factor. Regarding the user's rating sequence as a random discrete signal, and then randomly sorting the user's ratings k times, the average kurtosis of the user is obtained. And take the kurtosis value as the credibility of expert users. Through experiments on multiple datasets including MovieLens, Jester, Booking-Crossings, and Last.fm, we have proved the advancement and reliability of our method.

Serverless computing and Function-as-a-Service (FaaS) are programming paradigms that have many advantages for modern, distributed and highly modular applications. However, the process of transforming a legacy, monolithic application into a set of functions suitable for a FaaS environment can be a complex task. It may be questionable whether the obvious advantages received from such a transformation outweigh the effort and resources spent on it. In this paper we present our continuing research aimed at the transformation of legacy applications into the FaaS paradigm. Our test subject is an airport visibility system, a sub-class of the meteorological services required for airport operations. We have chosen to modularize the application, divide it into parts that can be implemented as functions in the FaaS paradigm, and provide it with a simple cloud-based data management layer. The tools that we are using are Apache OpenWhisk for FaaS and Airflow for workflow management, Apache Airflow for workflow management and NextCloud for cloud storage. Only a part of the original application has been transformed, but it already allows us to draw some conclusions and especially start forming a generalized picture of a Function-as-a-Service application.

Cloud computing performance optimization is the process of increasing the performance of cloud services at minimum cost, based on various features. In this paper, we present a new approach called MOOA-CSF (Multi-Objective Optimization Approach for Cloud Services Finding), which uses supervised learning and multi-criteria decision techniques to optimize price and performance in cloud computing. Our system uses an artificial neural network (ANN) to classify a set of cloud services. The inputs of the ANN are service features, and the classification results are three classes of cloud services: one that is favorable to the client, one that is favorable to the system, and one that is common between the client and system classes. The ELECTRE (ÉLimination Et Choix Traduisant la REalité) method is used to order the services of the three classes. We modified the genetic algorithm (GA) to make it adaptive to our system. Thus, the result of the GA is a hybrid cloud service that theoretically exists, but practically does not. To this end, we use similarity tests to calculate the level of similarity between the hybrid service and the other benefits in both classes. MOOA-CSF performance is evaluated using different scenarios. Simulation results prove the efficiency of our approach.

In order to effectively ensure the retrieval effect of enterprise data center resources, improve the retrieval accuracy of enterprise data center resources, and shorten the retrieval time of enterprise data center resources, a retrieval technology of enterprise data center resources based on density peak clustering algorithm is proposed. Analytical clustering algorithms, density clustering algorithms, and density peak clustering algorithms are all types of clustering algorithms. To reduce the dimensionality of enterprise data center resources, the kernel principal component analysis method is used. The structure of the enterprise data center resource set is reorganized and the feature quantity of the enterprise data center resource distribution is extracted using feature space reorganization technology. On this basis, the density peak clustering is carried out on the data center resource set of enterprise, and the semantic association distribution model of data center resource retrieval in enterprise is constructed. Through the semantic registration and weighted vector combination control method, the retrieval of enterprise data center resources is realized. The experimental results show that the proposed algorithm has a good effect on the retrieval of enterprise data center resources, which can effectively improve the resource retrieval accuracy and shorten the resource retrieval time.

In order to ensure the dynamic matching effect of human resources allocation and improve the accuracy and efficiency of dynamic matching of human resources allocation, a dynamic matching algorithm of human resources allocation based on big data mining is studied. Analyze the meaning and function of big data mining, and explain the common analysis principles of big data mining. The information entropy is selected as the basis for measuring human resource allocation, the human resource allocation is extracted, and the similarity of human resource allocation is calculated using the Huasdorff similarity method based on time interpolation. According to the Apriori algorithm and FP-Growth classification algorithm, the human resource allocation is classified and mined, and the K-Means clustering algorithm is used to realize the dynamic matching of human resource allocation. The experimental results show that the proposed algorithm has better dynamic matching effect of human resources allocation, and can effectively improve the accuracy and efficiency of dynamic matching of human resources allocation.

It remains a challenging task to identify human activities from a video sequence or still image due to factors such as backdrop clutter, fractional occlusion, and changes in scale, point of view, appearance, and lighting. Different appliances, as well as video surveillance systems, human-computer interfaces, and robots used to study human behavior, require different activity classification systems. A four-stage framework for recognizing human activities is proposed in the paper. As part of the initial stages of pre-processing, video-to-frame conversion and adaptive histogram equalization (AHE) are performed. Additionally, watershed segmentation is performed and, from the segmented images, local texton XOR patterns (LTXOR), motion boundary scale-invariant feature transforms (MoBSIFT) and bag of visual words (BoW) based features are extracted. The Bidirectional gated recurrent unit (Bi-GRU) and the Bidirectional long short-term memory (Bi-LSTM) classifiers are used to detect human activity. In addition, the combined decisions of the Bi-GRU and Bi-LSTM classifiers are further fused, and their accuracy levels are determined. With this Dempster-Shafer theory (DST) technique, it is more likely that the results obtained from the analysis are accurate. Various metrics are used to assess the effectiveness of the deployed approach.

Misogyny is characterized by hostility, hatred, aversion, intimidation, and violence against women. With the rise of social media, it has become one of the most convenient platforms for expressing woman-hating speech. As a result, misogyny is gaining appeal and societal standards are being violated. With millions of Bangladeshi Facebook users, misogyny is growing increasingly prevalent in Bangla as well. In this paper, we have proposed automatically identifying misogynistic content in Bangla on social media platforms in order to evaluate the problem's challenges. As there is no existing Bangla dataset for analyzing misogynistic text, we generated our own. We have applied various deep-learning algorithms to improve the classification of misogynistic text categories. LSTM and RNN models are used for designing the model architecture in deep learning. Models are evaluated using the confusion matrix, accuracy, and f1-scores. The results indicate that LSTM outperforms RNN in terms of accuracy by 67

In today's modern world, the usage of technology is unavoidable, and the rapid advances in the Internet and communication fields have resulted in the expansion of wireless sensor network (WSN) technology. However, WSN has been proven to be vulnerable to security breaches. The harsh and unattended deployment of these networks, combined with their constrained resources and the volume of data generated, introduces a major security concern. WSN applications are extremely critical, it is essential to build reliable solutions that involve fast and continuous mechanisms for online stream analysis, allowing the identification of attacks and intrusions. Our aim is to develop an intelligent and efficient intrusion detection system by applying an important machine learning concept known as ensemble learning in order to improve detection performance. Although ensemble models have been proven to be useful in offline learning, they have received less attention in streaming applications. In this paper, we examine the application of different homogeneous and heterogeneous online ensembles in sensory data analysis on a specialized WSN detection system (WSN-DS) dataset in order to classify four types of attacks: Blackhole attack, Grayhole, Flooding, and Scheduling among normal network traffic. Among the proposed novel online ensembles, both the heterogeneous ensemble consisting of an Adaptive Random Forest (ARF) combined with the Hoeffding Adaptive Tree (HAT) algorithm and the homogeneous ensemble HAT made up of 10 models achieved higher detection rates of 96.84