Tạp chí Khoa học Lạc Hồng

Camera-based security system featured with laser detection and warning

2025-03-11T06:49:19+00:00

Recent advancements in camera-based technologies have heightened security demands across various sectors. Cameras now play a pivotal role in traffic management, disease screening, industrial automation, and facial recognition. Nonetheless, many existing solutions are prohibitively expensive and complex, limiting the accessibility for broader applications. This paper introduces an innovative, low-cost, and compact security system utilizing the ESP32-CAM kit, which integrates a microcontroller with a built-in camera and Wi-Fi . Our system is engineered to detect unauthorized individuals in residential settings and parking facilities, employing a laser sensor to trigger alerts via a buzzer and capture photographic evidence for subsequent analysis. Anomalies identified by the system are seamlessly transmitted to cloud storage over Wi-Fi, ensuring real-time monitoring and response. Designed for easy installation and affordability, our system holds significant promise for enhancing security in homes and public areas. Through comprehensive experiments and evaluations, we validate the system's feasibility and effectiveness, demonstrating its potential as a transformative solution in the security landscape.

Malware detection in PE files using deep learning with self-supervised learning techniques

2025-02-12T04:05:00+00:00

In recent years, there has been a surge in new malware created by hackers globally, posing challenges for traditional detection methods. This paper explores using advanced artificial intelligence, specifically Deep Learning with Self-supervised learning, to identify malware in executable files. Our study focuses on comparing the effectiveness of popular deep learning techniques like CNN models and fine-tuned CNN models, against Autoencoder models. The key contribution of this paper lies in comparing the results of these different approaches to malware detection.

Unlocking the potential of blockchain in education: An innovative approach

2025-02-27T02:38:16+00:00

Modern technological advancements such as 5G networks, smart devices, and the Internet of Things (IoT) have established a new digital landscape where dataflow security is a mandatory factor. In this context, blockchain technology stands out as a promising solution with the ability to provide a secure and invulnerable platform for systems. Blockchain is a distributed ledger, not only serving financial transactions but also securely and sustainably storing any type of data. Although blockchain originated from cryptocurrency (Bitcoin), its applicability has expanded into various fields such as data storage, product certification, healthcare, science, and education. In the field of education, blockchain is used to issue and store certificates, support degree management, assess learning outcomes, maintain academic records, and oversee training processes. This article analyzes the features and benefits of blockchain, presenting its current applications in education. At the same time, it discusses the benefits and challenges of implementing this technology in the field of education.

Optimizing hidden layers in LSTM networks to generate detailed summaries of hotel reviews

2025-03-11T08:41:29+00:00

Online hotel booking platforms often lack the ability to provide detailed summaries of user reviews across key service areas, such as food, accommodation, service quality, and location. This study introduces a breakthrough solution using Long Short-Term Memory (LSTM) networks with optimized hidden layer configurations. With an F1- score of 75.28%—an increase of 10.18% compared to the standard LSTM model—the model has proven effective in generating review summaries for specific aspects. This advancement offers users a smarter and more efficient hotel booking experience while addressing the limitations of current review systems.

Modeling user success in online social networks using advanced GNN architectures

2025-02-13T02:49:47+00:00

Online social networks (OSNs) provide extensive data reflecting users’ personalities, interests, and social connections. The study explores how graph convolutional neural networks (GCNNs) can be used to analyze data from the VKontakte social network to predict users' professional success. Using features like user profiles and social connections, it evaluates various GCNN architectures, including GCNConv, SAGEConv, and GINConv. The Graph Isomorphism Network (GIN) layer achieved the highest accuracy (0.88). This research highlights the effectiveness of advanced neural networks in understanding professional success metrics in online social networks. Hong Bang International University, Ho Chi Minh City, Vietnam

Exploring tourist behavior using natural language processing: Evidence from tripadvisor reviews

2025-03-11T08:39:59+00:00

This study applies natural language processing (NLP) and Non-negative Matrix Factorization (NMF) to analyze TripAdvisor reviews of short-day tours in Southern Vietnam. Key topics representing traveler experiences and motivations are identified, framed within the push and pull theory. Push factors, like the desire for knowledge and convenience, and pull factors, such as unique activities and cultural immersion, influence tourist behaviors. The findings provide actionable recommendations for tour operators to create experiences that meet both intellectual and experiential needs, enhancing customer satisfaction and deepening connections with the destination.

Integrating pre-trained LLMS with RAG for efficient content retrieval

2025-03-11T08:21:13+00:00

Large Language Models (LLMs) are highly effective at replicating human tasks and boosting productivity but face challenges in accurate data extraction due to prioritizing fluency over factual precision. Researchers are addressing these limitations by combining LLMs with Retrieval-Augmented Generation (RAG) models. This approach utilizes chunking, searching, and ranking algorithms to streamline data retrieval from unstructured text, improving LLMs’ precision and processing. The findings provide key insights into optimizing chunking strategies and set the stage for the advancement and broader application of RAG-enhanced systems.

Applying AI technology to detect and rescue children stuck in cars

2025-02-20T01:34:50+00:00

Artificial Intelligence (AI) is a technology that enables computers to perform tasks that require human intelligence such as learning, reasoning, and problem solving. AI can be applied to solve serious problems such as children suffocating due to being left in cars. By using facial and voice recognition technology, AI can detect the presence of children in cars and sound a warning to the occupants and guide them to get out of the car. To achieve this, AI must be taught to recognize the presence of children through analyzing a large amount of image and sound data. Set the recognition sensitivity from 50% to 100% to activate the warning system. When the camera recognizes less than 50%, the system will not work. From 50 to 100%, the system will work. A 95% image recognition result will activate the warning speaker and turn on 2 relay switches. The noise recognition result of 80% will activate the warning speaker and the light on the board will change to red. When detecting someone, the PIR sensor will change the light color on the board to red. The recognition result will be continuously checked and adjusted to ensure accuracy. When the system detects a child in the car, it will immediately warn on the circuit board light and speaker. At the same time, it will send a message to the phone immediately. Thanks to the warnings’ high accuracy of up to 95%, this system can reduce the risk of accidents and thus contribute to the safety of children.

Overview of machine learning applications in manufacturing: From specialization to transfer learning

2025-03-11T08:30:11+00:00

The shift of manufacturing systems toward the paradigms of Industry 4.0 and 5.0 has significantly boosted the integration of Machine Learning (ML) technologies in this field. With the rapid growth of research leveraging ML to enhance manufacturing functions, this review aims to provide a thorough and up-to-date overview of these applications. A total of 114 journal articles were collected, analyzed, and categorized based on supervision approaches, ML algorithms, and application domains. The study highlights the benefits of ML in manufacturing, alongside identifying potential avenues for future research. Notably, the paper highlights that the prevalent focus on highly specialized applications in manufacturing could be mitigated by encouraging the implementation of transfer learning within the industry.

Improving facial expression recognition through PCA and LBP with SVM classifier

2025-02-12T04:08:46+00:00

This paper proposes a method for facial expression recognition method using Principal Component Analysis (PCA) and Local Binary Pattern (LBP) algorithms to extract feature from facial images. Experiments were conducted on two datasets: Japanese Female Facial Expression (JAFFE) database and Cohn-Kanade Extended (CK+) database. Support Vector Machine (SVM) was used as the primary classifier, compared to Euclidean distance (L2) to evaluate the performance of the classification methods. The experimental results show that the combination of PCA and SVM achieved a recognition rate of 87% on the JAFFE database and 81% on the CK+ database, with differences due to the complexity and diversity of the CK+ dataset. The study indicates that the PCA and LBP method combined with SVM outperforms methods using Euclidean distance, providing SVM to be an effective classifier for facial expression recognition in complex experimental environments.

Application of AI technology in safety monitoring and warning at secondary schools

2025-03-03T07:09:31+00:00

The research "application of ai technology in safety monitoring and warning at secondary schools" aims to develop a system for monitoring and ensuring railing safety in secondary schools by integrating Artificial Intelligence (AI) with IoT sensors. The system is designed to detect potential risks of students falling from railings and provide timely alerts to minimize accidents. The research methodology involves utilizing an AI camera for image recognition, a motion sensor to detect movement, and a distance sensor to measure the gap between individuals and the railing. Additionally, the system integrates a warning speaker and sends alert messages to mobile phones in hazardous situations. Data is collected and processed using an image recognition algorithm with an accuracy of over 80%, combined with sensor signals to determine safe or unsafe conditions. The motion sensor operates with an accuracy of 92%, while the distance sensor has an error margin of less than 5%. Experimental results indicate that the system achieves an image recognition accuracy of 85% in good lighting conditions and 78% in low-light environments. The findings demonstrate that the system effectively detects risks and issues timely warnings, paving the way for further research and applications in public spaces to enhance user safety.

Predicting employee attrition using machine learning approaches

2025-04-16T08:29:20+00:00

Employee attrition poses a critical challenge to organizations, both in terms of financial costs and operational continuity, with the average replacement cost per hire estimated at USD 4,129 and a reported attrition rate of 57.3% in 2021. This study applies machine learning techniques to predict employee attrition and identify its primary organizational drivers. Four supervised learning models were evaluated, Support Vector Machine (SVM), Support Vector Machine (LR), Decision Tree Classifier (DTC), and Extra Trees Classifier (ETC), in which the optimized ETC achieving the highest prediction accuracy of 93%, surpassing existing state-of-the-art methods. An Employee Exploratory Data Analysis (EEDA) revealed that monthly income, hourly rate, job level, and age are key factors influencing attrition. These findings highlight the effectiveness of AI-driven approaches in workforce analytics and provide actionable insights for organizational leaders aiming to improve retention through data-informed strategies.