IAES International Journal of Artificial Intelligence (IJ-AI)
Not a member yet
1769 research outputs found
Sort by
Recommendation system for football player recruitment using k-nearest neighbor
Get PDFIn modern professional football, achieving a competitive edge depends not only on on-field performance but also on effective off-field strategies, particularly in player recruitment. This study proposes a machine learning-based recommendation system to support talent identification and optimal player placement using statistical performance data. The model analyzes a wide range of features, including shots, expected goals, expected assists, pass types, offensive contributions, and defensive actions across field zones. The dataset undergoes preprocessing steps such as normalization (per 90 minutes) and dimensionality reduction. A key innovation of this research is the use of principal component analysis (PCA) to reduce feature dimensionality, minimizing redundancy while retaining essential information, which improves model efficiency and scalability. The refined data is then processed using the k-nearest neighbors (KNN) algorithm with cosine similarity, allowing the system to identify players with similar performance profiles based on directional similarity in a high-dimensional space. This combination enhances recommendation accuracy by focusing on performance structure rather than raw values. The resulting system provides actionable insights into player suitability and potential, offering clubs a data-driven tool for informed scouting and recruitment decisions. The approach demonstrates the effectiveness of combining PCA and KNN in optimizing football player recommendation systems
Enhancing emotion recognition model for a student engagement use case through transfer learning
Get PDFDistance education has been prevalent since the late 1800s, but its rapid expansion began in the late 1990s with the advent of the online technological revolution. Distance learning encompasses all forms of training conducted without the physical presence of learners or teachers. While this mode of education offers great flexibility and numerous advantages for both students and teachers, it also presents challenges such as reduced concentration and commitment from students, and difficulties in course supervision for teachers. This article aims to study student engagement on distance learning platforms by focusing on emotion detection. Leveraging various existing datasets, including the Facial Expression Recognition 2013 (FER2013), the Karolinska Directed Emotional Faces (KDEF), the extended Cohn-Kanade (CK+), and the Kyung Hee University Multimodal Facial Expression Database (KMU-FED), the proposed approach utilizes transfer learning. Specifically, it exploits the large number and diversity of images from datasets like FER2013, and the high-quality images from datasets like KDEF, CK+, and KMU-FED. The model can effectively learn and generalize emotional cues from varied sources by combining these datasets. This comprehensive method achieved a performance accuracy of 96.06%, demonstrating its potential to enhance understanding of student engagement in online learning environments
A novel ensemble-based approach for Windows malware detection
Get PDFThe exponential growth of internet-connected devices, particularly accelerated by the COVID-19 pandemic, has brought forth a critical global challenge: safeguarding the security of transmitted information. The integrity and functionality of these devices face significant threats from various forms of malware, leading to behavioral distortions. Consequently, a vital aspect of cybersecurity entails accurately identifying and classifying such malware, enabling the implementation of appropriate countermeasures. Existing literature has explored diverse approaches for malware identification, encompassing static and dynamic analysis techniques like signature-based, behavior-based, and heuristic-based methods. However, these approaches face a key issue of inadequately identifying unknown malware variants, often resulting in misclassifications of new strains as benign. To tackle this challenge, this study introduces a novel ensemble-based approach for identifying and classifying malware on Windows platforms, with a specific focus on detecting new and previously unknown variants. The proposed approach leverages multiple machine learning schemes to identify elusive unknown malware that proves challenging for existing methods.
An ontology-based knowledge modeling for the rite of Bai Sri Su Kwan: a ritual of the Greater Mekong Subregion
Get PDFThe development of ontologies is crucial in digital humanities research. This study focuses on creating a system to extract meaning from knowledge related to the Bai Sri Su Kwan ritual. Addressing semantic gaps, the second phase of our research outlines methods for developing an ontology for Bai Sri Su Kwan rituals. To fully understand this significant ritual in the Mekong Basin, we employed a theoretical framework with seven ontology development steps, using the Hozo Ontology Editor. Our ontology includes nine main classes: Bai Sri Su Kwan (A ritual subclass), persons, chants, belief, purpose, wish, literature, locations, and equipment. The Bai Sri Su Kwan subclass connects with all other classes in the ontology. This ontology forms the basis for a meaning search system for the Bai Sri Su Kwan ceremony in future research stages. The ontology was evaluated syntactically through human assessment and the OOPS! Ontology Pitfall Scanner. Validation results for the Bai Sri Su Kwan Ontology show no pitfalls in critical dimensions, indicating high integrity and reliability
Performance assessment of time series forecasting models for simple network management protocol-based hypervisor data
Get PDFTime series forecasting is vital for predicting trends based on historical data, enabling businesses to optimize decisions and operations. This paper evaluates forecasting models for predicting trends in simple network management protocol (SNMP)-based hypervisor data, essential for resource allocation in cloud data centers. Addressing non-stationary data and dynamic workloads, we use PyCaret to compare classical models like autoregressive integrated moving average (ARIMA) with advanced methods such as auto ARIMA. We assess 30 models on metrics including CPU utilization, memory usage, and disk reads, using synthetic and real-time datasets. Results show the naive forecaster model excels in CPU and disk read predictions, achieving low root mean squared errors (RMSE) of 0.71 and 869,403.35 for monthly and daily datasets. For memory usage predictions, gradient boosting with conditional deseasonalisation and detrending outperforms others, recording the lowest RMSE of 679,917.6 and mean absolute scaled error (MASE) of 4.46 on weekly datasets. Gradient boosting consistently improves accuracy across metrics and datasets, especially for complex patterns with seasonality and trends. These findings suggest integrating gradient boosting and naive forecaster models into cloud system architectures can enhance service quality and operational efficiency through improved predictive accuracy and resource management
SQL-CB-GuArd: a deep learning mechanism for structured query language injection attack detection
Get PDFStructured query language (SQL) injection attacks, which take advantage of input field vulnerabilities to introduce malicious code into database queries, are a serious danger to database-driven programs and systems. Intruders can now alter, recover, or remove sensitive data because of illegal access. Strong artificial intelligence (AI) based security solutions are required to reduce SQL injection threats, as these assaults' significance highlights. This study's main goal is to create automated AI-based techniques that can identify structured query language injection attack (SQLIA) in real time eliminating the need for human intervention. Although machine learning (ML) and deep learning-based techniques have received a lot of interest in this field, MLbased techniques have problems with accuracy and false negatives. Deep learning (DL) is therefore commonly used in these text data processing and natural language processing (NLP) applications. We have introduced a hybrid DL approach for SQLIA detection in this paper. The pre-processing step performs decoding, generalization, and tokenization to improve the learning performance. The proposed approach uses combination of convolutional neural network (CNN), bidirectional long short-term memory (Bi-LSTM), gated recurrent unit (GRU) with attention mechanism. The combination helps to improve the pattern learning capacity. The proposed approach is validated on publically available data and experimental analysis reported that the proposed SQL-CB-GuArd achieves better accuracy of SQLIA detection
Method for developing and partitioning graph-based data warehouses using association rules
Get PDFThe evolution of modern databases has led to a variety of not only structured query language (NoSQL) models, particularly graph-oriented-databases. This growth has encouraged businesses to explore graph-based business intelligence (BI) solutions. This paper explores three essential aspects in the domain of graph warehouse: the establishment of efficient graph warehouses, the significance of data historization, and the development of effective strategies for graph partitioning. It starts by building a BI system within a graph database. Subsequently, the paper emphasizes the pivotal role of data historization, highlighting the slowly graph changing dimension (SGCD) approach as a versatile framework for accommodating varied dimensional changes, additionally; the paper introduces a novel partitioning strategy utilizing association rules algorithms, for optimized and scalable graph warehouse management
Enhancing sepsis detection using feed-forward neural networks with hyperparameter tuning techniques
Get PDFThis paper investigates the use of feed-forward neural networks for sepsis detection, emphasizing class imbalance mitigation and hyperparameter optimization. Leveraging random oversampling, synthetic minority over-sampling technique (SMOTE), and random sampling techniques, we address class imbalance, significantly improving feed-forward neural network performance. The resulting model achieves an impressive 83% accuracy on the test set, with notable enhancements in precision, recall, and F1-score for the positive class. Hyperparameter tuning using RandomizedSearchCV identifies optimal parameters, including an alpha value of 0.01 and the logistic activation function, leading to a remarkable 57.5% test accuracy. GridSearchCV also contributes to model refinement, albeit with a slightly lower test accuracy of 51.5%. These findings underscore the importance of robust hyperparameter tuning methods in optimizing feed-forward neural network models for imbalanced datasets, particularly in sepsis detection. The insights gained hold promise for the development of more accurate diagnostic tools, ultimately improving patient outcomes in clinical practice
Evaluating ChatGPT’s Mandarin “yue” pronunciation system in language learning
Get PDFBy incorporating voice control technology into ChatGPT, it becomes possible to engage in conversations or dialogues with individuals who are actively engaged in the process of acquiring language skills. Our study team conducted a modest experiment to evaluate the efficacy of a voice control feedback system in facilitating the mastery of the most challenging pronunciation of the Mandarin syllable "yue". The objective of this study is to evaluate the effectiveness of voice-controlled ChatGPT in aiding learners to acquire accurate pronunciation of the Mandarin phoneme "yue". Furthermore, the study seeks to investigate the methods utilised by the ChatGPT model in identifying and distinguishing the word "yue" when it is used alone or in combination with "ye" and "yi". We employed many testing approaches, including single-word instances, paired instances, and the integration of phrases. In addition, we evaluated the model's ability to accurately detect the term "yue" in short sentences and, ultimately, in a longer sentence
Optimizing queue efficiency: Artificial intelligence-driven tandem queues with reneging
Get PDFThis paper delves into the theoretical integration of queueing theory and artificial intelligence (AI), examining the benefits and implications of their convergence. Queueing systems serve as fundamental models for various real-world applications, from telecommunications networks to healthcare facilities. This research presents a transformative framework for elevating the efficiency and performance of queueing systems by infusing AI-driven tandem queue analysis. The implications of this approach transcend industries, promising streamlined operations, reduced waiting times, and resource optimization. This work invites further exploration and application, offering a path to more effective and responsive queueing systems globally. Over the years, researchers and practitioners have explored numerous techniques to enhance the efficiency and performance of queueing systems. In recent times, integrating AI into the realm of queueing analysis has opened up new avenues for optimization and innovation. This paper studies a two-server tandem queueing model with reneging customers using AI techniques. Assuming that the arrival rate follows the Poisson process and the service rate follows an exponential distribution, using the birth-death process, probability generating function and AI module, we derive steady-state difference equation, expected number of people in customers, and mean waiting time