Computer Science and Application
ISSN / EISSN: 21618801 / 2161881X
Published by: Hans Publishers
Total articles ≅ 1,943
Latest articles in this journal
Computer Science and Application, Volume 13, pp 334-348; https://doi.org/10.12677/csa.2023.133032
In order to analyze the performance of P2P networks and solve the problem of excessive energy consumption of P2P network system. In this paper, the peers that resource requests in a P2P net-work are abstracted into customers and the peers that provide services are abstracted into servers. Based on the classical queueing model, two types of servers bulk service and asynchronous vacation strategies are introduced to establish the M/Md/c+k(0≤k≤d) queueing model. The steady-state distribution of the system is solved by using the matrix-geometric solution method, and then the performance indicators such as the average queue length and the average waiting time of the system are derived. This paper also focuses on the energy consumption of the system in different periods, and puts forward the Nash equilibrium and the social optimal strategy to find appropriate parameters to improve the problem of high energy consumption in P2P networks.
Computer Science and Application, Volume 13, pp 349-357; https://doi.org/10.12677/csa.2023.133033
In recent years, with the increasing number of investors, traders frequently buy and sell products with high market volatility, such as gold, and bitcoin is used to maximize profits. This article studies the use of historical price data of currencies in the trading market to avoid risks in the investment process, predict the direction of product prices, and obtain maximum returns. For investors, if they can have a good trading strategy, this will bring them a stable income, while also ensuring the smooth operation and stability of the trading market. In order to better predict the currency trend, this paper uses historical data to predict the next day’s currency price to construct LSTM, random forest, gradient recovery tree and xgboost model for single-step prediction, through comparison, the random forest model has the best prediction effect. This paper uses the prediction results to propose a method for portfolio optimization based on dynamic programming, which considers the impact of risk factors on the portfolio management process, can automatically convert the portfolio optimization mode according to market conditions and asset information to cope with market style changes, and adjust the portfolio asset composition and asset allocation in real time through the dynamic trading of portfolio internal assets and external asset pools, so as to maximize theoretical profits.
Computer Science and Application, Volume 13, pp 301-310; https://doi.org/10.12677/csa.2023.133029
Due to the large changes in facial features, the correct recognition rate of the original face is low. In view of the phenomenon, this experiment proposed a hybrid self-attention block structure for rec-ognizing faces with facial features changes. For this reason, 26 kinds of micro-plastic surgery small sample image data sets were made by ourselves. Integrating self-attention into the bottleneck block of the residual network improves the ability of the hybrid self-attention block to capture the features of each region of the image. The experiment on the small sample micro-plastic data sets shows that the hybrid self-attention network proposed in this experiment has a higher correct recognition rate: 89.70%, the correct recognition rate increased by 2.65% compared with ResNet50, and the correct recognition rate of the hybrid selfattention model with improved connection increased by 1.12% compared with the hybrid self-attention model without improved connection, and the net-work performance was also improved.
Computer Science and Application, Volume 13, pp 311-318; https://doi.org/10.12677/csa.2023.133030
With the development of science and technology and information technology, “information over-load” makes it difficult for users to quickly and accurately obtain the content they want from a large number of movie resources, which has become a major constraint affecting user experience. There-fore, personalized film recommendation has become the focus of research in the new era. The un-precedented development of deep learning has achieved great success in many fields. In order to alleviate the problem of data sparsity, this paper integrates multiple attribute information into the feature representation of users and items to obtain a complete primary feature representation of users and items. In addition, attention mechanism is used to distinguish the importance of these attributes, and then two fusion strategies, splicing and outer product, are used to model the poten-tial features of user movie, and multi-layer perceptrons and convolutional neural networks are respectively used to fully learn the nonlinear interaction between users and movies. Experiments on the movie dataset MovieLens 1M show that compared with the traditional DeepCF model, the model proposed in this paper improves the hit rate by 4.19% and the normalized cumulative loss gain by 0.38% respectively.
Computer Science and Application, Volume 13, pp 319-333; https://doi.org/10.12677/csa.2023.133031
For a highly distributed heterogeneous environment such as mobile P2P network, in order to better achieve the goal of cooperation and resource sharing of various terminal systems, maintain and improve the performance of the network, building a network topology with good performance is the key to research, in view of this feature, based on the hybrid mobile network model combining semi-distributed topology and fully distributed topology, a two-stage queuing model with retry space and working fault is established. The probability distribution under steady-state of the net-work model is obtained by using the mimic-life and extinction process, the matrix geometric solution method and the Gauss-Seidel iterative method, and the expression of performance indicators such as the average number of nodes in the two stages is given. The numerical experiment is carried out by programming software, the influence of parameters on each index is analyzed, and the optimal parameters are solved by constructing the average cost function per unit time and the social utility function of the system, which provides a decision-making basis for the mobile P2P net-work model.
Computer Science and Application, Volume 13, pp 270-280; https://doi.org/10.12677/csa.2023.132027
In order to realize automatic garbage detection and grasping work and solve the inaccuracy and labor of the workers operating the refuse collection point, this paper uses a real-time garbage detection algorithm based on the YOLOv7 algorithm, which can detect the garbage in each partition by taking real-time photos through the camera in the garbage pool area and grasp the garbage to be grasped into the area to be incinerated in real-time through the crane jaws, effectively improving the production efficiency. The experimental results show that the algorithm of this paper has a mAP value of 87.7% in 100 rounds of training, which can better meet the needs of real garbage in real-time industrial processing.
Computer Science and Application, Volume 13, pp 281-300; https://doi.org/10.12677/csa.2023.132028
This paper introduces a service portfolio revenue optimization algorithm based on evolutionary game theory. In the traditional edge process automation control system, it is common to make algorithm decisions that require centralized optimization and flexible adjustment of output. Usually, the decision-making process of these algorithms needs to be separately configured and solved in specific software system modules. Its onsite operability is not friendly to engineers, and in the process control process, the calculation modules that rely on algorithms are usually configured in the cloud central computing unit. The results usually need to be transmitted to the edge control equipment by field communication. The real-time reliability of the algorithm processing is limited by the stability of the communication system, which has potential threats to the accuracy and stability of the edge process automation control system. Based on the consideration of the above problems, this paper proposes a self-organizing intelligent algorithm for component-based resource services that can run in low computational power runtime. Through the analysis of evolutionary stability strategy, the evolutionary path of cooperative behavior between edge resources is discussed. The evolutionary game is encapsulated by components through the function block structure conforming to IEC 61499 standard, and the reusability and real-time operability of algorithm calculation in the edge process control system are improved by components.
Computer Science and Application, Volume 13, pp 251-258; https://doi.org/10.12677/csa.2023.132025
With the rise of AI (Artificial Intelligence), many industries have become closely related to AI, while making autonomous driving a reality step by step. The coexistence of human-driven vehicles with autonomous vehicles is a necessary stage to achieve fully autonomous driving. In this paper, we study the expected impact on urban road traffic flow due to the implementation of different char-acteristics and penetration rates of smart driving vehicles. This paper establishes a model describing intelligent and human-driven vehicles and uses adaptive priority algorithms to investigate the different state evolution processes of human-machine driven hybrid traffic flows, laying the foundation for the study of the theory and methodology of hybrid traffic flows. The study found that under different weather conditions, with the different penetration rate distribution of autonomous vehicles, the traffic and environment of mixed traffic flow were positively affected to a certain extent, and the overall running time was optimized by 7.68% when the penetration rate of intelligent vehicles was increased to about 50%. Under the condition of snow and ice, when the penetration rate of intelligent driving vehicle increases to 30%, the optimization rate of running time reaches the highest value of 5.01%. This shows that the increasing number of smart driving vehicles will help reduce traffic congestion.
Computer Science and Application, Volume 13, pp 259-269; https://doi.org/10.12677/csa.2023.132026
In the field of binocular stereo vision, stereo matching is an important research direction. In order to solve the problem that some stereo matching algorithms have high error matching rate in the weak texture region, this paper proposes a stereo matching algorithm based on multi-scale and multi feature. The STAD, gradient and improved Census cost fusion are used as the cost computing method. In the cost aggregation stage, take the guided filtering algorithm as the core. The cost cubes of different scales are fused using the idea of cross scales, and different cost aggregation parameters are set for the cost cubes of different scales. For some errors in disparity map results, a variety of methods of disparity post-processing are used. The experimental results show the accuracy of the algorithm in the weak texture area. The experimental results of standard image pairs on the Mid-dlebury 3.0 test platform show that the average mismatch rate of the algorithm in multiple groups of weak texture images is 8.16%, which has higher matching accuracy than the traditional SGM and other algorithms.
Computer Science and Application, Volume 13, pp 236-250; https://doi.org/10.12677/csa.2023.132024
Bioinformatics is a science that integrates advanced biological science and computer technology. It integrates mathematics, information science and computer technology to scientifically organize, sort out and conclude the information of biology and medicine. DNA sequence alignment is one of the most important and basic research directions in bioinformatics and an important means to explore the relationship between genes and diseases. The main objective of this paper is to find all sequences that are identical to the target sequence and whose occurrence probability is greater than the given threshold in the uncertain molecular sequence data and to give the total number of target sequences and the starting site of each target sequence. In this paper, a weighted suffix tree-based DNA sequence pattern matching algorithm is proposed to solve the problem that the existing molecular sequence pattern matching algorithm based on “space for time” is limited to the calculation of times, and the image stereo matching method based on the double DNA sequence alignment algorithm in bioinformatics is limited to uncertain source data. This method uses weighted suffix trees as the main data structure, improves the matching accuracy of uncertain source data, and solves the problem that map data structure is limited to number calculation. Experimental results show that the proposed algorithm has improved the matching speed and sensitivity to a certain extent.