Articles

Access the latest knowledge in applied science, electrical engineering, computer science and information technology, education, and health.

Advanced Search
Filter Icon

Filters article

Years

FAQ Arrow
Single Year
Range Year
0
0

Source Title

FAQ Arrow

Authors

FAQ Arrow

30,096 Article Results

GAN-augmented vision transformer with balanced synthetic data generation for robust rice leaf disease detection

10.11591/ijece.v16i3.pp1307-1318
Saiful Islam , Md. Nasim Akhtar , M. Mahadi Hassan , A. N. M. Rezaul Karim , Israt Binteh Habib
Early and accurate identification of rice leaf diseases is essential for sustainable crop management; however, many existing convolutional neural networks (CNNs) based solutions struggle with class imbalance and limited robustness when applied to real-field data. In this work, a generative adversarial network (GAN) augmented vision transformer (ViT) framework is introduced to overcome these limitations. A deep size representative samples for underrepresented disease categories, resulting in a more balanced training dataset and achieving a Fréchet inception distance (FID) score of 18.6. The balanced dataset is then used to train a vision transformer model that leverages self-attention to capture global contextual features of rice leaf images. Experimental evaluation across ten disease classes shows that the proposed approach attains an overall classification accuracy of 96.5%, exceeding the performance of several established CNN architectures. Additionally, the model demonstrates strong generalization capability on an external field dataset, achieving 94.8% accuracy. To validate real-world applicability, the trained model is deployed on a Jetson Nano edge device, where it delivers efficient inference performance suitable for practical agricultural applications. The findings indicate that combining GAN-based data augmentation with transformer-based learning provides a reliable and scalable solution for rice leaf disease detection.
Volume: 16
Issue: 3
Page: 1307-1318
Publish at: 2026-06-01

Retrieval-augmented generation in enterprise knowledge systems: architecture, benefits, and applications

10.11591/ijece.v16i3.pp1407-1416
Mohammad Baqar
This paper presents an adaptive retrieval-augmented generation (RAG) framework for enterprise knowledge systems that combines multi-source ingestion, semantic indexing with Hugging Face embeddings and Facebook artificial intelligence similarity search (FAISS), metadata-aware retrieval, and grounded large language model generation. The research addresses a persistent enterprise gap: critical knowledge is distributed across documentation, tickets, code repositories, and collaboration tools, while static keyword search and periodically retrained language models cannot keep pace with rapidly changing operational data. The proposed approach contributes a privacy-preserving architecture, a retrieval-and-feedback loop that improves ranking quality over time, and a unified workflow that links evidence retrieval to solution recommendation. In an evaluation over a 1.2 million-document corpus and a six-week pilot, the framework improved Precision@10 from 0.58 to 0.81, reduced documentation retrieval latency from 45.6 s to 12.3 s, and shortened average bug-resolution time from 18.4 h to 7.2 h. These findings indicate that enterprise RAG can materially improve troubleshooting speed, knowledge reuse, and decision support while maintaining stronger control over sensitive organizational data. The broader implication is that adaptive, governed RAG systems can serve as a practical foundation for future enterprise artificial intelligence (AI) assistants, analytics platforms, and compliance-aware decision workflows.
Volume: 16
Issue: 3
Page: 1407-1416
Publish at: 2026-06-01

A risk-constrained SARSA–FIS hybrid decision architecture with adaptive exploration control

10.11591/ijece.v16i3.pp1531-1542
Joni Fat , Parwadi Moengin , Pudji Astuti , Sally Cahyati
Algorithmic trading systems operate in highly dynamic and uncertain environments where learning-based decision agents must balance adaptability with strict risk control. Reinforcement learning (RL) methods provide adaptive policy optimization but often suffer from unstable exploration and limited interpretability in financial markets. This study proposes a risk-constrained SARSA–FIS hybrid decision architecture with adaptive exploration control for algorithmic trading. The framework integrates a compact SARSA-based reinforcement learning environment with a Sugeno-type fuzzy inference system (FIS) that converts reinforcement signals into interpretable trading decisions. Exploration follows a decaying ε-greedy policy with a drawdown-triggered reset mechanism to maintain bounded risk exposure during learning. The system was implemented as a MetaTrader 5 Expert Advisor and evaluated on the GBPUSD currency pair using historical market data. Experimental results show that the hybrid framework improves trading performance compared with a rule-based baseline. During a six-month out-of-sample evaluation, the system achieved a net profit of 90 USD and a profit factor of 1.35, compared with 10 USD and 1.02 for the baseline. Extended one-year testing confirmed stable profitability and controlled drawdown behavior. The results demonstrate that integrating reinforcement learning, fuzzy decision mapping, and explicit risk constraints provides a practical approach for developing adaptive trading agents.
Volume: 16
Issue: 3
Page: 1531-1542
Publish at: 2026-06-01

Optimization of transfer learning for facial emotion classification on the FER-2013 dataset

10.11591/ijece.v16i3.pp1213-1226
Nida Muhliya Barkah , Shofwatul ‘Uyun
Facial expressions play a key role in non-verbal communication by naturally reflecting human emotions. Facial emotion recognition (FER) using computer vision has gained attention with advances in deep learning. However, deep learning models require large datasets to perform well, posing a challenge for FER tasks with limited data. Transfer learning is a promising approach to address this issue, but a standardized method for FER is yet to be established. This study optimizes three transfer learning models ResNet-50, Inception V3, and Xception on the FER-2013 dataset. Experiments include testing input image sizes, hyperparameter tuning, data augmentation, layer addition, and training methods. Results show each model requires different input sizes for best accuracy. Hyperparameter tuning improves accuracy by 6.35%, 4.69%, and 1.04% for ResNet-50, Inception V3, and Xception, respectively. Augmenting only the disgust class yields better accuracy than augmenting all classes. The freeze fine-tuning method is less effective than fine-tuning alone on datasets with thousands of samples but outperforms the freeze layer method. The best accuracies achieved are 64.89% (ResNet-50), 65.83% (Xception), and 66.40% (Inception V3). These findings provide insights into freeze fine-tuning limitations and guidance for optimizing transfer learning in FER with limited data.
Volume: 16
Issue: 3
Page: 1213-1226
Publish at: 2026-06-01

Designing and evaluating a community-based digital dictionary system for the Balinese language: An IT innovation adoption study

10.11591/ijece.v16i3.pp1369-1381
Cokorda Pramartha , Madek Jeani Purnama , Ida Bagus Gede Sarasvananda , I Wayan Arka , Ni Luh Watiniasih
Regional and vulnerable languages increasingly depend on digital tools to remain visible and usable in everyday life, yet many dictionary initiatives are described mainly in terms of content or interface features rather than evaluated as information-system innovations. This paper presents an exploratory design science study of a community-based Balinese digital dictionary that supports bidirectional Balinese-Indonesian lookup, Latin and Balinese Unicode script, speech-level information, part-of-speech tagging, related-word search, and role-based contribution workflows. The platform is implemented as a web-based system with a three-tier architecture and relational database. To evaluate adoption readiness, 40 users completed representative tasks and then responded to an adapted Moore and Benbasat IT innovation adoption instrument covering seven constructs. The results show high ease of use, relative advantage, and compatibility, indicating strong functional value and fit with user routines. Image and visibility are moderate, while result demonstrability and visibility show lower reliability and are therefore interpreted as exploratory indicators. The study contributes both a documented digital-dictionary artefact for Balinese language support and a reusable evaluation approach for other early-stage community-facing information and communication technology (ICT) systems. The findings suggest that wider uptake depends not only on technical quality, but also on institutional visibility, outreach, and continued content enrichment.
Volume: 16
Issue: 3
Page: 1369-1381
Publish at: 2026-06-01

Analyzing learners' perceptions of engagement and learning interaction in gamified massive open online courses for TVET using SEM-PLS

10.11591/ijece.v16i3.pp1319-1328
Azizul Mohd Yusoff , Sazilah Salam , Siti Nurul Mahfuzah Mohamad , Rujianto Eko Saputro
The introduction of gamified massive open online courses (G-MOOCs) represents a novel advancement in technical and vocational education and training (TVET). The use of gamification in education has been shown to increase engagement and motivation, which are crucial for effective learning. However, there is limited research on the specific impacts of G-MOOCs on learner outcomes in TVET. A key feature of G-MOOCs is the integration of gamification elements to enhance learner engagement and interest. This research employs structural equation modelling with partial least squares (SEM-PLS) to examine learners' perceptions of their participation and learning experiences in G-MOOCs for TVET. Specifically, the study aims to identify how gamification approaches such as fun, engagement, and learner interaction influence knowledge acquisition, skills development, satisfaction, and overall learning outcomes. The analysis reveals that G-MOOCs have a strong positive correlation (0.505) with learning engagement. Additionally, learning engagement significantly moderates learning outcomes (p=0.002). Interaction also has a significant impact (p=0.381) on learning outcomes. Overall, the findings indicate a significant positive relationship between learners' activities and their performance in G-MOOCs.
Volume: 16
Issue: 3
Page: 1319-1328
Publish at: 2026-06-01

Hybrid deep learning (ILeS-Net) for lung cancer classification in cloud-IoT healthcare systems

10.11591/ijece.v16i3.pp1588-1607
Affrose Affrose , Cheruku Sandesh Kumar , Archek Praveen Kumar
This study presents a cloud–Internet of Things (cloud-IoT) based intelligent decision support framework for lung cancer classification and treatment recommendation, centered on a hybrid deep learning model termed ILeS-Net. Computed tomography (CT) images from a benchmark dataset are first preprocessed using Gaussian filtering to enhance image quality. Cancerous regions are identified using an Improved BIRCH (I-BIRCH) segmentation approach, followed by feature extraction using shape descriptors, color features, and Improved local Gabor XOR pattern (I-LGXP) textures. The extracted features are classified using ILeS-Net, which integrates Improved LeNet-5 and SqueezeNet architectures to achieve improved classification performance with reduced overfitting. Based on the classification results, the framework provides supportive recommendations to assist clinical decision-making. Experimental results demonstrate that the proposed ILeS-Net model achieves a maximum accuracy of 0.951, outperforming several conventional and state-of-the-art methods. The cloud–IoT integration further enables scalable, real-time, and secure data processing, highlighting the framework’s potential for practical computer-aided lung cancer diagnosis.
Volume: 16
Issue: 3
Page: 1588-1607
Publish at: 2026-06-01

An internet of things-telemedicine platform empowered by 5G mobile networks for Tunisian Rural places

10.11591/ijece.v16i3.pp1261-1271
Ibrahim Monia , Dadi Mohamed Bechir , Rhaimi Belgacem Chibani
With the advent of Internet of Things (IoT) technologies, offering new possibilities for remote healthcare delivery, the medicine sector has undergone significant advancements in recent years. New tools are used, and diagnostics have become more accurate. We suggest creating a platform that can be extended for several applications. This platform has been realized to attest and demonstrate how IoT technology offers devices that could be integrated to provide novel services like remote consultations. Our proposed platform contains novel functionalities such as real-time video calls, instantaneous messaging, live notifications, vital signs monitoring, and electronic health record access. This is accomplished with enhanced qualities of remote healthcare services. Added to this, healthcare access equity will be guaranteed. The paper emphasizes the potential of Laravel 11 as a framework offering powerful features for creating modern and high-performance applications. We have integrated Laravel Reverb, a powerful real-time communication package, to provide seamless real-time communication with users. With our application, notifications and interactions are dynamically created. This allows instant updates to delivery and engages the user experience. The database was designed based on the latest version of MySQL 8, coupled with the advanced capabilities of PHP 8.2. This combination provides unparalleled performance, scalability and reliability. Added to that, IoT’s technology usage helps to improve healthcare access and delivery, especially in underserved areas. Human and machine cooperation is a main factor of the 5th industry level. This is widely respected by our platform. This offers great help, especially for those isolated and underserved areas, as we hope.
Volume: 16
Issue: 3
Page: 1261-1271
Publish at: 2026-06-01

Design and implementation of NMPC for a two-DOF robotic arm using CasADi

10.11591/ijra.v15i2.pp307-318
Lahcen Boulbalah , Faiza Dib , Nabil Benaya , Khaddouj Ben Meziane
Achieving accurate joint-space tracking in multi-link robotic arms is complicated by strong configuration-dependent nonlinearities and mandatory actuator limits that classical controllers are structurally unable to enforce. This paper presents a nonlinear model predictive control (NMPC) scheme for a two-degree-of-freedom (2-DOF) serial robotic arm, implemented within the CasADi symbolic computing environment to leverage automatic differentiation and sparse interior-point solving. The complete set of Lagrangian equations of motion-inertia, Coriolis, and gravity terms-is incorporated directly into the optimizer's prediction model through fourth-order Runge-Kutta (RK4) integration, eliminating the need for linearization. Torque, velocity, and angle bounds are imposed as native hard inequality constraints at every step of the finite-horizon optimization. Systematic simulations pit the proposed NMPC against a Ziegler-Nichols-tuned decentralized PID at two distinct sampling periods. The NMPC achieved a 95% reduction in peak tracking error relative to PID (0.0058 rad vs. 0.1347 rad for Joint 1), with mean error decreases of 64.65% and 57.58% for Joints 1 and 2 respectively, at an average solver time of 0.053 s-comfortably within the 0.1 s control cycle. The findings demonstrate that online NMPC with unabridged nonlinear dynamics is computationally practical for real-time joint control on standard computing hardware.
Volume: 15
Issue: 2
Page: 307-318
Publish at: 2026-06-01

Design of beefsteak tomato harvesting robot system in greenhouse

10.11591/ijra.v15i2.pp353-364
Thien An Dinh , So Nam Phung , Tri Cong Phung
One challenge for tomato harvesting robots is that some of the tomato stems were not detectable because they were hidden behind the leaves or other obstacles. The primary objective of this research is to design, simulate, and experiment with a tomato harvesting robot and propose an improved detection algorithm to overcome the above problem. The suggested detection algorithm is designed to first detect the tomato fruit itself, and if the stem is not visible, the system will automatically adjust the camera's viewing angle to provide a better perspective and uncover the hidden stem. Simulation and experimental tests were carried out in a real tomato greenhouse to evaluate the cutting and holding mechanism, as well as the camera-based detection algorithm. These experimental results confirmed the effectiveness of the gripper and detection system and revealed several challenges in the harvesting algorithm. By integrating advanced algorithms for tomato detection and harvesting, this robot will reduce damage to the tomatoes, ensuring higher quality and yield.
Volume: 15
Issue: 2
Page: 353-364
Publish at: 2026-06-01

Machine learning-driven analysis of user bandwidth allocation and performance in 5G heterogeneous network: a survey

10.11591/ijece.v16i3.pp1236-1248
Pang Wai Leong , Raymond Chia , Phang Swee King , Goh Hui Hwang , Chan Kah Yoong , Chung Gwo Chin
A key foundation of 5G heterogeneous networks (HetNets) is the use of network slicing, which divides bandwidth into multiple logical networks and accounts for each function’s requirements. Currently, various machine learning (ML) models are being implemented into the network slicing algorithm to allocate bandwidth dynamically. The network slicing algorithm analyzes the traffic and allocates bandwidth based on the current services using a network-centric approach. However, limited work is found on further studying the impact of user-centric algorithms in bandwidth allocation. This paper presents the network slicing used in 5G and the limitations of these algorithms. A detailed review of user-centric bandwidth allocation algorithms is presented, along with a critical review of ML algorithms for traffic prediction and resource allocation decisions. Finally, the technology gaps and opportunities of the existing works are reported, and the direction for further research of ML in user-centric bandwidth allocation algorithms is tabulated.
Volume: 16
Issue: 3
Page: 1236-1248
Publish at: 2026-06-01

Energy-aware inertial measurement units scheduling for wearable LoRa systems using quaternion features

10.11591/ijece.v16i3.pp1449-1465
Yudhi Adhitya , Indri Septiani
Wearable Internet of Things systems increasingly depend on inertial measurement units (IMUs) to capture human motion, yet continuous high-frequency sensing, on-device processing, and long-range (LoRa) communication impose significant energy and latency challenges for battery-powered devices. This study formulates a practical scheduling framework that optimizes IMU sampling, quaternion-based feature extraction, and transmission decisions within the wearable/LoRa architecture. The framework operates in discrete time windows of W=0.5−1 s, within which sensing, processing, and communication decisions are updated at the window level to balance energy consumption and responsiveness. The method models energy consumption, accuracy degradation at lower sampling rates, and communication constraints to define feasible operating modes and determine optimal configurations under varying activity levels. An empirical accuracy–frequency mapping and component-wise energy model support both offline optimization and lightweight online scheduling. The results show that the proposed framework can balance accuracy, responsiveness, and battery life by dynamically shifting between high-performance, balanced, and low-power surveillance states. This scheduling strategy extends operational lifetime while preserving motion-detection reliability and ensuring timely event transmission. The findings demonstrate the importance of energy-aware IMU management in long-range wearable systems and provide a foundation for adaptive sensing strategies in real-world deployments.
Volume: 16
Issue: 3
Page: 1449-1465
Publish at: 2026-06-01

A multi-modal framework for improving the accuracy of phishing email detection

10.11591/ijece.v16i3.pp1608-1625
Lamees Mohamed Faraj , Sayed Abdel-Gaber , Hanan Fahmy
Phishing emails continue to pose a significant cybersecurity threat, particularly through the increasing use of malicious attachments to evade traditional text-based detection systems. Most existing approaches focus primarily on email content, creating a blind spot in attachment-aware phishing detection. This paper proposes a multi-modal phishing email classification model that integrates email header features, body text analysis, and attachment inspection within an ensemble learning framework. Independent machine learning classifiers are employed for each email component, and a majority voting mechanism is used to determine the final classification decision. The proposed model is evaluated using publicly available email and attachment datasets that are combined to simulate attachment-bearing phishing emails. Experimental results demonstrate strong detection performance across multiple evaluation metrics. Nevertheless, the study acknowledges the limitation of using synthetically paired email bodies and attachments, which may not fully capture real-world semantic relationships. The findings highlight the importance of incorporating attachment-aware analysis into phishing detection systems and provide a foundation for future research on semantic consistency modeling and transformer-based architectures.
Volume: 16
Issue: 3
Page: 1608-1625
Publish at: 2026-06-01

An enhancement of stock price forecasting based on hybrid BiLSTM-Transformer model

10.11591/ijece.v16i3.pp1298-1306
Pham Hoang Vuong , Lam Hung Phu , Le Nhat Duy , Pham The Bao , Tan Dat Trinh
Stock price forecasting presents a challenging problem due to factors like nonlinearity, seasonality, and economic volatility in financial data. Deep learning approaches can handle nonlinearity and complexity of financial data, but they often face limitations in capturing both local and global dependencies. This study introduces a hybrid Transformer–bidirectional long short-term memory (BiLSTM) model to improve stock price forecasting. Our method combines the strength of BiLSTM with the global context understanding of the Transformer by embedding a 1D convolutional layer. The model can efficiently capture short-term and long-term dependencies in stock data. Experimental results on various datasets show that our hybrid model outperforms other well-known models.
Volume: 16
Issue: 3
Page: 1298-1306
Publish at: 2026-06-01

A critical review of information retrieval techniques: current trends and challenges

10.11591/ijict.v15i2.pp456-464
Sanket D. Patil , Zahir Aalam
The realm of information retrieval is witnessing transformative advancements, driven by the integration of deep learning techniques, specialized algorithms, and domain-specific applications. Information retrieval systems play an important role in many applications including in the Artificial Intelligence powered systems that can be seen in many applications. Information Retrieval, generally, acts an important task in the knowledge discovery phase of any query based intelligent system. This paper presents a comprehensive review by conducting a detailed analysis of the technological nuances, dataset specifications, and pivotal findings. This detailed review has been done with the special emphasis on the kind of technology used to achieve accurate information retrieval, domain of the study, and the system’s ability to retain or work with tables and figures, among other parameters. Navigating through the rich tapestry of methodologies, the paper underscores the pivotal role of deep learning frameworks in revolutionizing traditional retrieval paradigms. Furthermore, it sheds light on the innovative integration of textual information, algorithmic advancements, and specialized datasets to enhance the efficacy and granularity of information retrieval mechanisms.
Volume: 15
Issue: 2
Page: 456-464
Publish at: 2026-06-01
1 2 3
...
2006 2007
Show 7 of 2007

Discover Our Library

Embark on a journey through our expansive collection of articles and let curiosity lead your path to innovation.

Explore Now
Library 3D Ilustration