https://doi.org/10.5573/IEIESPC.2025.14.1.1

(Kumaravel Kaliaperumal) ; (Meenakshi L. Rathod) ; (Leo Raju) ; (J. Mahi)

In social and real-time network applications, community detection is the very popular and rapidly expanding field of study. Recently, many community detection approaches have been developed. In instance, community detection has proved to be effective and successful in local development strategies. Nevertheless, there are few basic problems to expose the overlapping communities. Although certain techniques are not sensitive enough to demonstrate widespread overlaps, the maximal approaches allow the seeds to be initialized and parameters to be created. A new unsupervised Map Reduce dependent local expanding technique to overlap community dependent seed node finding is presented in this study. The proposed method finds the leader or seed nodes of communities by using simple graph metrics, including closeness, centrality, degree, and betweenness. It then finds the communities that follow the leader nodes. Map-Reduce-oriented is proposed to utilize the Fuzzy C-Means Clustering approach to find which communities overlap depending on the leader nodes. The experimental outcomes illustrate the proposed method (LBCD), which assess network graph allowed the overlapping community structures, is more effectual and confident when used to entire 11 actual-world data sets.

https://doi.org/10.5573/IEIESPC.2025.14.1.11

(Peng Xiao)

The lack of interactive guidance in the art self-learning process is currently the main problem that learners face, which often makes it difficult to continue learning. To solve this problem, the study designs an interactive art teaching system. This system is based on the iOS system and is primarily aimed at art beginners. The instructional system and learners participate in educational exchanges using the iOS painting image style classification model. Further, the system realizes interaction between students through the painting description generation model. This process facilitates intra-platform interaction. The results shows that the classification accuracy of the designed painting image style classification model is above 80.00%, and the average recall rate is 78.16%. In addition, the designed painting description generation models are accurate in the range of 84.45% to 97.74%, and the recall rate is in the range of 83.71% to 98.27%. The study demonstrates the effectiveness of the interactive art teaching system for providing guidance to beginner in the art department.

https://doi.org/10.5573/IEIESPC.2025.14.1.22

(Ziwei Cui)

Expression recognition is a crucial aspect of emotion computing research, with significant value in distance education, human-computer interaction, and medical research. This study analyzes students’ emotions on an art education platform with a focus on facial expression recognition algorithms using static images. To overcome the limitation of single facial features, a feature fusion algorithm is proposed to weigh and fuse regional features. The improved Local Binary Patterns (LBP) algorithm enhances texture feature extraction by utilizing a new threshold to describe image pixel relationships. Additionally, the Histogram of Oriented Gradients (HOG) extracts edge features from the eyebrow and mouth areas, which are then fused with the texture feature of the face using optimal weighting coefficients. The fused features are classified and recognized using Support Vector Machine (SVM) classifiers. Comparative experiments on the JAFFE and CK+ datasets demonstrated that the fusion feature-based facial expression recognition algorithm outperformed individual algorithms, achieving a recognition rate improvement of 12.7% and 13.6%, respectively. These findings offer theoretical insights and support for emotional analysis in the context of students’ art education.

https://doi.org/10.5573/IEIESPC.2025.14.1.33

(Xiuqin Jiang) ; (Jianbin Zhen)

The continuous improvement of internet technology has promoted the application of big data prediction models in employment prediction, providing more diversified solutions for research and analysis in this field. This paper presents an employment prediction model that combines association rules with an optimized RBF neural network. This model is designed to predict and analyze academic performance and employment situations more efficiently and accurately. By comparing the performance of different models, it can be seen that the employment prediction model constructed in this article has smaller prediction errors; In addition, the analysis of the impact of different impact projects on the most employment rate and the correlation between each project also indicates that there are certain differences in the impact of different projects on the employment rate, and the degree of correlation between each impact project also varies to a certain extent. The model constructed in this article can achieve higher accuracy, accuracy, and sufficient reliability, providing new ideas and research methods for predicting and analyzing academic performance and employment rates in the field of education.

https://doi.org/10.5573/IEIESPC.2025.14.1.45

(Xiaodong Li)

With the rapid development of computer networks, the types and quantities of network failures are also significantly increasing. To apply fault-tolerant recognition to the detection of network interruption faults, a small-scale mathematical model for fault-tolerant recognition of network interruption faults is constructed. Firstly, small-scale network interrupts are identified. Then, fault-tolerant recognition is used to mathematically model network interrupts. Finally, simulation experiments are used to verify the performance of the mathematical model. The results showed that the successful running rates of the constructed mathematical model in random mode and hot mode were 89.63% and 81.95%, respectively. Under the same ratio, the average utilization rates in random mode and hotspot mode are 87.53% and 88.21%, respectively. This indicates that under the same conditions, the model has a high resource utilization rate, which can better complete data transmission. This verifies the application effect of the fault-tolerant recognition mathematical model in small-scale network interruption faults, aiming to provide a new research direction for fault-tolerant recognition of network interruption faults.

https://doi.org/10.5573/IEIESPC.2025.14.1.57

(Shan Tong) ; (Yuting Zhang) ; (Shaokang Li)

The powerful feature learning capability of the AlexNet model of convolutional neural networks is gradually applied in land classification, but the model suffers from problems such as inconsistency between sample form and network requirements in its application. Therefore, this study proposes the optimized AlexNe-finetunet model, LCNet-27 model and LCNet-13 model based on AlexNet model to solve the problems of oversized training samples and model training input size, speed up model training and improve classification accuracy. The highest accuracy was 97.76% for the LCNet-27 model and 95.33% for the LCNet-13 model with 5×5 pixel input size, both higher than that of the AlexNet model finetune, 93.6% of the AlexNet model finetune. The LCNet-27 and LCNet-13 models obtained by optimizing the AlexNet model finetune with this study have improved the accuracy and faster classification speed in remote sensing image land classification, solving the problem of contradiction between sample size and model input size of the AlexNet model, which can be well applied in land classification problem. The novelty of this work lies in the innovative use of relative models for different studies for multiple aspects of land classification problems, using the AlexNe-refinement model, the LCNet-27 model, and the LCNet-13 model for a more comprehensive land classification study.

https://doi.org/10.5573/IEIESPC.2025.14.1.68

(Zhiqi Fan)

Traditional remote virtual teaching lacks emotional interaction like face-to-face teaching, affecting learning and teaching effectiveness in distance education. In view of this, the study innovatively introduces a self-cure neural network on the basis of convolutional network. Then, calibration strategies, regularization sorting, and noise labeling operations are used to optimize the network threshold, proposing a new facial expression recognition model. In addition, emotional space and emotional transfer pathways are constructed, and hidden Markov models, forward backward algorithms, and motivational factors are introduced to propose a motivational interaction model for emotional regression. The experimental results showed that the highest recognition accuracy of the expression recognition model was 95.8%. The recognition error was the lowest at 33% when the label noise ratio was 70%. The lowest misidentification rate was 34% when the obstruction proportion was 67%. The average emotional intensity of the incentive interaction model in multiple environments was 0.074. The average compensation time for multiple incentive factors was 13 minutes, which was 6 minutes shorter than that of a single incentive factor. The above results indicate that the proposed model can achieve accurate facial expression recognition, providing technical support for emotional interaction between teachers and students.

https://doi.org/10.5573/IEIESPC.2025.14.1.83

(Young-Lak Kim, Chun-Hyung Cho)

With the rapid development of computer network technology and virtual reality technology, VR display system has been widely used because of its unique display way. Based on B/S architecture, this paper proposes a web-based interactive VR scene collaborative display system for interior design. Firstly, B/S architecture is used to realize the sharing and interaction of cooperation instructions between client and server, and XML data transmission format and Ajax polling mechanism are used to realize the network sharing and transmission of 3D model data. Starting from virtual reality and network database technology, a set of VR display system based on Web is developed by using virtual reality modeling language VRML, Java, MYSOL and ASP. The application server is implemented by the Java language, and the client is a Java Applet embedded in the HTML page, a VRML reader plug-in and a part of the HTML page. Applets and VRML readers communicate via an External Authoring Interface (EAI). Explore the image stitching principle of the Mosaic panoramic camera, and put forward the realization scheme of spherical panoramic stitching on the mobile end, using the gyroscope sensor rotation Angle main code and video rendering, and finally combined with the multi-channel video stream synchronous decoding algorithm based on time stamp to realize the panoramic visualization experience based on the mobile end. System performance test shows that the depth map calculation time of 12 sets of panoramic pictures of the interior design interactive VR system proposed in this paper ranges from 10-15s, and the average calculation time is 12.19s, which meets the requirements of the system.

https://doi.org/10.5573/IEIESPC.2025.14.1.96

(Yuzhe Shen)

Current multi-objective algorithms face challenges in optimizing old city renovations within the context of smart city development due to the risk of getting trapped in local optima. To address this issue, a study developed an indicator system that integrates the Analytic Hierarchy Process with standards’ importance determination through assessing their correlation for weight evaluation. The particle swarm optimization algorithm was introduced to optimize the transformation plan, adjusting the inertia weight of particles, optimizing the particle swarm algorithm, and conducting simulation verification. Findings indicated a rapid surge in the number of overloaded nodes and pipe sections when the rainfall’s return period exceeded 5 years, illustrating the limited capacity of built sponge facilities to handle rainstorms. Under 50-year conditions, node and pipeline overload rates peaked at 31.70% and 50.19%, respectively. In combined facility simulations, runoff control rates increased to 82.55%-85.55% and 31.86%-36.97% under varying rainfall intensities. The proposed multi-objective optimization approach successfully defined optimal deployment ratios for four facilities, enhancing response capabilities under extreme rainfall conditions. This method demonstrates efficacy for transforming and optimizing old urban areas, especially in sponge city redevelopment efforts.

https://doi.org/10.5573/IEIESPC.2025.14.1.109

(Muhammad Adnan Khan) ; (Iftikhar Naseer) ; (Muhammad Nadeem Ali) ; (Byung-Seo Kim)

The novel Coronavirus (COVID-19) spread rapidly around the world and caused overwhelming effects on the health and economy of the world. It first appeared in Wuhan city of China and was declared a pandemic by the World Health Organization (WHO). Many researchers, as well as experts in clinical and artificial intelligence experts, are working together to control the rapid spread of COVID-19 with early detection. This study focused on intelligent prediction for coronavirus using computational intelligence approaches (IPC-FCIA) like convolutional neural networks, support vector machines, and fuzzy logic techniques. The proposed IPC-FCIA model is based on two sections namely the training section and the validation section. Features fusion and decision-level fusion are used in this study to enhance the performance of the recommended IPC-FCIA model. The proposed model predicts the early detection of COVID-19 in two types COVID-negative and COVID-positive. The benchmark results of the model show an accuracy of 97.66 % on decision-level fusion in the detection of COVID-19. The proposed model can be helpful for medical experts as well as COVID-19-affected patients.

https://doi.org/10.5573/IEIESPC.2025.14.1.118

(Muhammad Adnan Khan) ; (Abdur Rehman) ; (Sagheer Abbas) ; (Muhammad Nadeem Ali) ; (Byung-Seo Kim)

The current COVID-19 epidemic is responsible for causing a catastrophe on a global scale due to its risky spread. The community’s insecurity is growing as a result of a lack of appropriate remedial measures and immunization against the disease. In this case, social distancing is thought to be an effective barrier against the spread of the contagion virus as the risk of virus transmission can be reduced by avoiding direct contact with people. Thus, the goal of this research is to develop and improve an AI (Artificial Intelligence) system architecture for social distance monitoring. The framework could also use the GPS (Global Positioning System) to recognize human separation through cell phones. The transition learning framework is also applied to improve the consistency of the existing system. In this manner, the detection system uses a pre-trained technique that takes a Bluetooth dataset and location-sharing dataset to link to an additional level. In an attempt to approximate social distancing breaches among people, we used Bluetooth technology along with GPS distance estimation and set a threshold. To predict if the distance value exceeds the required social distance standard, a violation threshold is calculated and then it sends an alarm to every individual who is not maintaining social distancing. In response, the individual who breaks the social distance limit is also monitored using a detection approach.

https://doi.org/10.5573/IEIESPC.2025.14.1.128

(Wen Pei) ; (Wen-An Pan) ; (Jui-Chan Huang)

In resource allocation decisions in business, fully understanding customers’ needs and preferences helps to maximise benefits. As a result, in the modern business environment, the design of customized recommendation systems has gained a lot of attention. To this end, the study designs a recommendation algorithm for resource allocation decision based on improved two-part graph network structure. In this algorithm, an improved K-means clustering algorithm is introduced to deeply mine potential information. The calculation of similarity between users is also optimised to assist the target user to find the real neighbouring users. The findings demonstrate that, in comparison to the other algorithms, the resource allocation recommendation algorithm based on improved bipartite graph suggested in the study has a greater hit rate. The hit rate of the suggested algorithm can reach 32.5% when the recommendation list length is 10, which is a 21.5% improvement over the collaborative filtering algorithm. The suggested algorithm’s popularity is only 39.1, which is 69.3 less than the collaborative filtering algorithm when the suggestion list length is 10. Furthermore, the suggested algorithm for resource allocation decision-making created by the research has a greater recommendation accuracy, more personalization, and diversity, as seen by the proposed algorithm’s mean Hamming distance of 0.976. Through an improved bipartite graph network, the algorithm can fully analyze the historical preference information of users, effectively capture the complex relationship between users and products, generate personalized recommendation lists, and improve user satisfaction and purchase conversion rates. It provides an effective recommendation role for resource allocation decisions in modern business and helps to create greater economic benefits.