https://doi.org/10.5573/IEIESPC.2024.13.5.425

(C. P. Shirley) ; (Depaa RA B) ; (A. Priya) ; (R. Sarala) ; (Rajdeep Singh Solanki) ; (Malini K. V) ; (Ch. Venkatakrishna Reddy)

This manuscript presents a hybrid method for optimal energy management in smart home appliances. The proposed approach combines the Ebola Optimization Search Algorithm (EOSA) with the performance of spiking neural networks (SNN). The key objective of the proposed strategy is cost reduction through day-ahead load scheduling while also considering the best demand response (DR), and self-consumption of photovoltaic (PV). The EOSA method manages air conditioning (AC) and ensures thermal comfort. In addition, the SNN method was used to predict the optimal system control signal. Comparative analysis with existing systems, such as PSO, RFA, and BCO, showed that the proposed strategy resulted in 13% cost savings. The proposed system considered parameters, such as solar radiation, occupant presence, and electricity prices. The proposed technique is implemented in the MATLAB software, and the performance is compared with existing approaches. The efficiency of the proposed approach is higher than the existing techniques.

https://doi.org/10.5573/IEIESPC.2024.13.5.435

(Ravi Kumar Suggala) ; (Suma Bharathi. M) ; (P.L.V.D. Ravi Kumar) ; (NVS. Pavan Kumar)

A decentralized computing system distributed among data-generating hardware and the cloud is called fog computing (FC). The ability to place resources to improve performance is given to users by such a flexible structure. On the other hand, low-delay services and limited resources make it difficult to use new virtualization technologies for task scheduling with resource management in fog computing. Several studies have examined scheduling and load balancing (LB) in cloud computing (CC). Nevertheless, countless LB initiatives have been proposed in fog environments. Task scheduling using a Dual interactive Wasserstein generative adversarial network (DIWGAN) optimized with an Improved Dwarf Mongoose Optimization algorithm is proposed to classify the suitable and not suitable server for the process (ESA-WF-FE-DIWGAN). Initially, the 'Cloud-Fog Computing Dataset is used. Afterward, the dataset is fed to the Fog Resource Monitor (FRM). Here, the statistical features like storage, computing, and RAM are extracted. Subsequently, the extracted features are given to DIWGAN to classify the suitable server and the unsuitable server for the process. The scheduling process was done using the Improved Dwarf Mongoose Optimization algorithm. The proposed approach achieves 3.101%, which was a 7.12% higher make span: 24.13% and 13.04% lower total cost; 2.292% and 5.365% higher ARU compared to the existing methods.

https://doi.org/10.5573/IEIESPC.2024.13.5.443

(Seonghyun Park) ; (Taeyoung Lee) ; (Jongsik Ahn) ; (Haemoon Kim) ; (Hyunhak Kim) ; (Seoyoung Kim) ; (Byungin Choi)

Infrared images are known to capture the thermal radiation emitted from objects and are increasingly essential in Night Vision and surveillance. These can be utilized in various image processing algorithms, such as object detection and tracking. However, infrared image processing is highly complex due to the sensor degradation and the status of temperature inversion between the background and the object, which results in an inadequate dataset. Data augmentation approaches have been introduced to overcome the lack of datasets by increasing the diversity of data distribution. Withal, the augmentation approach via image processing algorithms is widely used to improve model performance, prevent overfitting caused by insufficient data, and mitigate data bias. Furthermore, several recent studies have established novel algorithms to overcome dataset shortage and uniform distribution through domain shifts such as image generation and image-to-image translation. In this paper, the object detection performance with infrared data augmentation based on the diffusion models of "Palette" and "BBDM" are analyzed and evaluated from various perspectives, such as the number of images, class, and object size. The evaluation showed that the compound dataset of Palette and BBDM at the ratio of 20% and 10%, respectively, improved by 0.3% and 0.5% compared to the baseline. Nevertheless, the similar distribution of real and translated infrared images showed better qualitative and quantitative performances.

https://doi.org/10.5573/IEIESPC.2024.13.5.452

(Leping Ma)

Sensors have a wider range of applications in the medical field, and it can be said that most medical detection instruments are composed of sensors as the core. This article focuses on the innovative application of medical intelligent sensor technology, especially the research on electrocardiogram sensors that combine microelectronics technology and Raman spectroscopy technology. Its core function is to convert human physiological parameters into analyzable electrical signals. Meanwhile, in order to improve detection accuracy, this article introduces Raman spectroscopy non-destructive testing technology and optimizes the application of sensors in electrocardiogram monitoring. By comparing and analyzing with traditional detection methods, the unique advantages and technological changes of Raman spectroscopy technology in the field of non-destructive testing were explored. The research results indicate that it not only effectively solves the problem of data accuracy in medical testing, but also achieves precise monitoring of equipment, with high economic value and clinical application potential.

https://doi.org/10.5573/IEIESPC.2024.13.5.462

(Jing Pu) ; (Yuke Li)

In the time of boost of the Internet, online education is also thriving, but how to grasp the performance of students in the context of online education has become a major problem in the current online education. In response to the above problems, the study proposes the Principal Component Analysis Algorithm (PCA), Adaptive Network-based Fuzzy Inference System (ANFIS) and Genetic Algorithm (GA) for secondary school students' performance prediction model, i.e., PCA-GA-ANFIS model, and the performance of PCA-GA-ANFIS, ANFIS and XGBoost were analyzed and compared through experiments. The experiment illustrated that the loss value of PCA-GA-ANFIS model was about 0.15, while the loss values of ANFIS model and XGBoost model were about 0.3 and 0.25, respectively, which were higher than PCA-GA-ANFIS model. The absolute and relative errors of ANFIS model and XGBoost model were about 3.6, 2% and 3.2, 1.8%, respectively; the absolute and relative errors of the PCA-GA-ANFIS model are about 2.2 and 1.6%, respectively, and the minimum absolute error tends to be close to 0, which demonstrated that the forecasting outcomes of the PCA-GA-ANFIS model were closer to the true values. The PCA-GA-ANFIS model performed with an average accuracy of 89.4% and precision of 90.3%. The results demonstrated that the PCA-GA-ANFIS model outperformed the other two models in terms of accuracy and precision. Comparatively, the ANFIS and XGBoost models exhibited an average accuracy and precision of approximately 86.8% and 87.8%, and 87.0% and 89.1%, respectively. Therefore, the results showed that the PCA-GA-ANFIS model produced more precise prediction results and better prediction performance than the ANFIS and XGBoost models.

https://doi.org/10.5573/IEIESPC.2024.13.5.472

(Zhenzhen Tang)

A novel RPA technique for financial data analysis based on quantum neural network model is innovatively developed. Compared to the conventional model, the Hybrid Quantum Convolutional Neural Network model (HQCNN) reduces the loss value by 0.0056% and improves the training and testing accuracy by 0.93% and 0.36%, respectively. In addition, the hybrid quantum convolutional neural network model of this RPA exhibits less fluctuation in loss value after 60 iterations and remains stable after 90 iterations. Meanwhile, we combine the particle swarm optimization algorithm with HQCNN to transform classical data into quantum state data for joint optimization. The results of the recognition and validation accuracy plot show that PSO-HQCNN has the best recognition effect, with a recognition accuracy of 96.85%, followed by HQCNN with a recognition accuracy of 96.20%. And CNN has the lowest recognition accuracy of 95.65%. However, PSO-HQCNN model increases the time cost while improving the accuracy. Therefore, in the process of applying RPA to financial systems, trade-offs should be made according to the actual situation, and the HQCNN model or PSO-HQCNN model should be selected. In conclusion, through the implementation process of RPA robot visualization application established by using the quantum neural network model, this study provides a new direction of exploration and a feasible implementation path for the automated process processing of enterprise financial data analysis.

https://doi.org/10.5573/IEIESPC.2024.13.5.480

(Yihao Hou) ; (Zongzhe Lin)

Music is a powerful way to express emotions, and as information visualization develops, visualizing emotions in music has become a popular topic. This study proposes a strategy for visualizing emotions in music on mobile devices. It uses the activation-degree-effectiveness emotion model and combines the residual phase with mel-frequency cepstral coefficient weighting to extract emotion features. Convolutional and recurrent neural networks were optimized and used together to recognize musical emotions. Experimental results show that the proposed method achieves the highest recognition accuracy of 90% and 92% in the Sound-track dataset and Song’s dataset, respectively, and an error rate of 10% in the AMG1608 dataset. The accuracy for recognizing happiness, sadness, relaxation, and anger is above 88%. This study provides a feasible direction for optimizing the visual design of expression of emotion in music art and recognition of emotion in music.

https://doi.org/10.5573/IEIESPC.2024.13.5.490

(Haifen Li)

The application of visualization technology in education helps to promote information technology in education and enhances learners’ ability to perceive information. Therefore, this study builds an online, self-directed learning visualization teaching system that includes a data input layer, a visualization coding layer, a logical recommendation layer, and a view presentation layer. In addition, to enhance visualization of web data, the teaching system optimizes the visual experience by using the U-Net algorithm and an immersive network with a 3D layout. The results show that for the LIME dataset test, this study’s proposed algorithm obtained the highest mean image entropy at 7.67, which is an improvement of 0.05 and 0.32 over the MBLLEN and RetinexNet algorithms, respectively. Therefore, this algorithm is more advantageous in enhancing the richness of image information. In addition, in the validation of the rationality of the online self-directed learning model, the visualization technique was fully supported by students, helping them to adapt to the learning system quickly. This indicates that the visualization system proposed in this study is effective in improving students’ self-directed learning skills.

https://doi.org/10.5573/IEIESPC.2024.13.5.499

(Jibeom Kim) ; (Eun-Sung Jung)

In modern IT environments, the importance of disaster recovery in providing stable services arising from system failures or data loss is emphasized. Furthermore, with the emergence of various disaster recovery tools, their selection and utilization has become a core task for stable business operations. In this study, we conducted an in-depth investigation and analysis of disaster recovery tools in the Kubernetes environment, a widely used open-source platform for application deployment and management in companies. In particular, we investigate two examples of disaster recovery tools, “Backup and Restoration tools” and “Recovery tools in Multicluster Management Platforms.” The main characteristics and differences, such as recovery support, cloud connectivity, backup location support, Role-Based Access Control support, encryption support, user interface, cost, and automatic recovery features provided by each tool, were compared, and analyzed. Based on this, we examine the importance of disaster recovery in the modern IT environment and explore optimization strategies through the integration of backup and restoration features using various tools and platforms. Our study suggests the feasibility of developing an integrated system that allows automatic recovery at both cluster and application levels. This opens the door to future research and implementation and is expected to provide even more extensive and in-depth knowledge and experience.

https://doi.org/10.5573/IEIESPC.2024.13.5.514

(Muhammad Bilal Sarwar) ; (Ghulam Musa Raza) ; (Muhammad Ali Sarwar) ; (Byung-Seo Kim)

Artificial Intelligence (AI) and Machine Learning (ML) are rapidly transforming the Information and Communication Technology (ICT) sector. In this research study, we explore the potential of AI and ML in ICT. We begin by providing an overview of the fundamental principles of AI and ML. We then discuss the various applications of these technologies in ICT, highlighting their transformative potential. We also examine the challenges and ethical implications of AI and ML adoption. We argue that these technologies hold vast potential to catalyze significant transformations within the ICT sector and beyond. However, they also present new challenges and ethical considerations, demanding continuous research, robust regulatory frameworks, and a multidisciplinary approach to ensure their beneficial and ethical application. Our research presents an all-encompassing view of the remarkable potential of AI and ML to reshape the landscape of ICT. However, it also serves as a cautionary note on the importance of mindful, ethical, and responsible implementation of these powerful technologies. It is our shared responsibility to navigate this thrilling frontier in a way that ensures advancements lead to a future that is fair, sustainable, and inclusive for all.

https://doi.org/10.5573/IEIESPC.2024.13.5.523

(Hanwen Li) ; (Ben Jin)

Intelligent science and technology are constantly changing, and society and people’s lives are inseparable from science and technology, affecting the transformation of artistic creation. This study constructed a choreography model based on a mixed-density network and action feature filtering strategy to help solve the disadvantages of traditional choreography. The model combines the long short-term memory (LSTM) network and mixed-density network to generate dance movements. First, the LSTM gating mechanism was used to learn the characteristics of human dance movements and obtain the conversion rules of various poses. The mixed-density network was introduced to compensate for the uncontrollable probability distribution in LSTM. In addition, during action generation, the experiment focused on the continuity between adjacent actions to screen the generated dance actions according to the diversity of dance actions and to enhance the continuity and authenticity of dance. Finally, an experiment was conducted on the spatial feature extraction and music feature matching of the model to achieve the goal that the model can generate different styles of choreography. The test user gave a score of more than four points to the final choreography effects of different styles, showing that the model can achieve a better intelligent choreography effect.

https://doi.org/10.5573/IEIESPC.2024.13.5.534

(Jaewon An) ; (Sang Ho Choi)

Face-recognition using deep-learning techniques typically face problems with learning about angles and with flexibility. In this study, we propose a face recognition system specialized for angles by using gray images. The K-face dataset was used to perform learning, along with triplet loss to achieve angle-specific learning. We reconstructed the model structure of Inceptionresnetv2, a model with excellent face recognition performance and the proposed model achieved an accuracy of approximately 96.9 %. In addition, the performance verification for each illuminance and angle was conducted and showed that the proposed model was flexible to the angle and illuminance. To confirm the applicability of model in the real environment, 21 subjects were recruited and verified. Transfer learning was performed to ensure the flexibility of the model in the real environment and gray images were used to reduce the effect of illuminance. Consequently, we obtained approximately 81% of the results and it was demonstrated that the proposed face recognition system can be applied in a real environment.