cover
Contact Name
Heri Nurdiyanto
Contact Email
jurnal.ijasca@gmail.com
Phone
+6285766661199
Journal Mail Official
jurnal.ijasca@gmail.com
Editorial Address
Lucky Arya Residence 2 No. 18 Jalan HOS. Cokroaminoto Kab. Pringsewu 35373
Location
Kab. pringsewu,
Lampung
INDONESIA
International Journal of Advanced Science and Computer Applications
Published by UK Institute
ISSN : 28097599     EISSN : 28097467     DOI : https://doi.org/10.47679/ijasca
International Journal of Advanced Science and Computer Applications (IJASCA) is a peer-reviewed open-access journal. The journal invites scientists and engineers throughout the world to exchange and disseminate theoretical and practice-oriented the whole spectrum of Advanced Science and Computer Applications. Submitted papers must be written in English for an initial review stage by editors and further review process by a minimum of two international reviewers. Accepted papers will be freely accessed in this website
Articles 65 Documents
Integrated a combined-statistical technique of Importance-Performance Analysis (IPA) and Machine Learning algorithms to Enhance heart disease prediction Saeid Aghasoleymani Najafabadi; Saeid Jafarzadeh Ghoushchi; Elnaz Osgooei
International Journal of Advanced Science and Computer Applications Vol. 5 No. 2 (2026): September
Publisher : Utan Kayu Publishins

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47679/ijasca.v5i2.183

Abstract

Heart disease has been identified as one of the most common diseases among people worldwide, and early prediction of this disease can significantly help reduce its incidence. Selecting the most important influential parameters to reduce the risk of this disease is another vital factor in this field. Over time, with the advancement of science and technology, advanced methods, including artificial intelligence algorithms, have been an appropriate alternative to some of the old statistical and laboratory methods. These artificial intelligence algorithms, which include machine learning and deep learning architectures, can assist researchers in predicting heart diseases and identifying influential factors with high accuracy. For this purpose, in this study, which aims to predict heart disease and identify the most important influencing factors, a hybrid model that is the result of combining six machine learning and deep learning architectures, including Logistic Regression, Quadratic Discriminant Analysis (QDA), ANN, XGBoost, Naïve Bayes, and Gradient Boosting, was introduced on a combination of four datasets: Cleveland, Hungary, Switzerland, and VA Long Beach. The most important features were selected using the statistical feature analysis method, and the most important influencing variables were selected using the Importance-Performance Analysis (IPA) technique. The results of this study show that the proposed hybrid model, using the most important features, achieved the highest evaluation parameters, including an accuracy of 91%, recall of 92%, precision of 88%, and F1-score equal to 90%, and predicted healthy people from unhealthy people.
AI-Enhanced Threat Intelligence for Zero-Trust Cloud Infrastructures: A Cyber-Resilience Engineering Approach Md Imran Khan
International Journal of Advanced Science and Computer Applications Vol. 5 No. 2 (2026): September
Publisher : Utan Kayu Publishins

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47679/ijasca.v5i2.260

Abstract

Due to the rapid pace of cloud adoption, the business environment has shifted to scalable, flexible, and economical features of IT resources. The shift to cloud computing supports new cyber security challenges from attack vectors evolving beyond perimeter-based defense mechanisms. The distributed and elastic nature of cloud infrastructures greatly increases the attack surface which adds pressure on continuous assurance and adaptive protection mechanisms.The article describes an AI-Enabled Threat Intelligence Framework that can support security assurance to Zero-Trust Cloud Infrastructures (ZTCI) using real-time detection, analysis, and mitigation of security threats. The framework utilizes machine learning-enabled Automated policy privilege access and ongoing authentication. The suggested architecture performs unsupervised and deep learning to detect the unusual behavior of users and applies reinforcement learning to the dynamically evolving access policies according to their risk scores. The experiments conducted report the results with the help of the standard datasets UNSW-NB15 and CICIDS2017.It is shown that using the traditional rule-based detection systems, the detection performance improvement, the number of false alarms being decreased, and the response time being shorter were all statistically significant. The findings indicate that Ai powered analytics integration into the Zero Trust security approach makes it possible to be less vulnerable to zero-day vulnerabilities, insider attacks, and threats in multi-cloud environments. The work gives more understanding about smart, self-learning, and flexible security solutions and proposes a new security infrastructure with a flexible and anticipative design that fits the current cloud architecture.
LMM-Net: A Label-Masked Multi-Task Network with temporal-contour fusion for singing voice quality assessment long Gao
International Journal of Advanced Science and Computer Applications Vol. 5 No. 2 (2026): September
Publisher : Utan Kayu Publishins

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47679/ijasca.v5i2.268

Abstract

Automatic singing voice quality assessment is essential for benchmarking modern singing voice generation and conversion systems, yet subjective listening tests are costly and difficult to scale. This paper presents LMM-Net, a label-masked multi-task network that predicts multi-dimensional perceptual scores under incomplete supervision and emphasizes system-level ranking consistency. LMM-Net combines a self-supervised acoustic encoder with an explicit temporal pitch-contour branch. A gated fusion module injects contour cues in a reliability-aware manner, and attention pooling aggregates frame-level evidence into utterance-level representations for multi-task regression. Experiments on the SingMOS-Pro benchmark under the standardized 16 kHz protocol show that LMM-Net improves both utterance-level prediction and system-level ranking. The proposed model achieves a system-level LCC of 0.89 and a system-level SRCC of 0.87 with a system-level RMSE of 0.08, outperforming strong self-supervised baselines augmented with pitch statistics. Ablation studies further confirm that label masking, temporal-contour modeling, gated fusion, and end-to-end adaptation are all necessary to obtain robust gains, especially for system ranking. These results suggest that combining incomplete-label learning with structured temporal melodic modeling is a practical and effective direction for perceptual assessment of singing voice systems.
An Explainable and Reproducible Lightweight Machine Learning Framework for SMS Spam Detection Using TF-IDF and Linear Support Vector Classification Pradyumna Mishra; Akash Yadav
International Journal of Advanced Science and Computer Applications Vol. 5 No. 2 (2026): September
Publisher : Utan Kayu Publishins

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47679/ijasca.v5i2.269

Abstract

The continued use of short message service (SMS) for personal, institutional and transactional communication has made mobile messaging a persistent target for unsolicited advertising, social engineering and fraud-oriented content. This paper presents a reproducible and explainable lightweight machine learning framework for binary SMS spam detection. The framework integrates a fixed text-preprocessing sequence, term frequency-inverse document frequency (TF-IDF) representation, comparative classifier evaluation, ablation analysis and post-hoc model interpretation. Five supervised learning models were examined on the SMS Spam Collection dataset: Multinomial Naive Bayes, Logistic Regression, Linear Support Vector Classification, Random Forest and XGBoost. The dataset contained 5,572 labelled messages, including 4,825 ham and 747 spam instances, and was evaluated using a stratified 80:20 train-test split with a fixed random seed. The strongest verified configuration used unigram and bigram TF-IDF features without stop-word removal and a Linear Support Vector Classifier, producing 98.57% accuracy, 97.84% precision, 91.28% recall and 94.44% F1-score. Unlike studies that report only aggregate accuracy, this work analyses the precision-recall trade-off, confusion-matrix behaviour and representative error cases. Explainability is incorporated through SHAP LinearExplainer applied to the LinearSVC decision-function margin, supported by feature-weight inspection. The results indicate that a carefully configured sparse-text linear classifier remains a competitive, transparent and computationally modest baseline for SMS spam filtering, although broader validation on contemporary multilingual and adversarial message streams is necessary before real-world deployment.
A Cloud Computing and Big Data Analytics-Powered Intelligent University Student Management System: A Cloud Computing and Big Data Analytics Juan Zhou
International Journal of Advanced Science and Computer Applications Vol. 5 No. 2 (2026): September
Publisher : Utan Kayu Publishins

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47679/ijasca.v5i2.270

Abstract

The rapid expansion of higher education and the growing complexity of student data necessitate intelligent and scalable student management systems. Conventional approaches often struggle to identify students at academic risk, classify performance accurately, and process high-dimensional educational datasets efficiently. This study proposes an intelligent student management framework integrating big data analytics, cloud computing, an autoencoder, and an artificial neural network (ANN). Academic performance data are comprehensively collected and preprocessed through cleaning, transformation, and normalization. An autoencoder performs unsupervised feature learning and dimensionality reduction, extracting meaningful latent representations from complex student datasets. These features are subsequently supplied to an ANN classifier to categorize students according to pass/fail status, performance levels, and academic risk, enabling timely intervention and support. A cloud-based distributed computing architecture ensures scalable storage, flexible access, and efficient processing of large-scale educational data. Experimental results demonstrate that the hybrid autoencoder-ANN model achieves 97.83% accuracy, 97.65% precision, 97.83% recall, and 97.58% F1-score, confirming its strong predictive capability. Cloud evaluation records response times of 12.5–30 ms, latency of 8–20 ms, and throughput of 75,000–95,000 records per second. Overall, the proposed framework supports efficient, accurate, scalable, and data-driven academic decision-making and strengthens early intervention strategies for improving student success.