Williams Kyei^*, Chunyong Yin, Kelvin Amos Nicodemas, Khagendra Darlami

Journal on Artificial Intelligence, Vol.8, pp. 19-38, 2026, DOI:10.32604/jai.2026.074988 - 20 January 2026

Abstract The COVID-19 pandemic has underscored the need for rapid and accurate diagnostic tools to differentiate respiratory infections from normal cases using chest X-rays (CXRs). Manual interpretation of CXRs is time-consuming and prone to errors, particularly in distinguishing COVID-19 from viral pneumonia. This research addresses these challenges by proposing a customized EfficientNet-B0 model for ternary classification (COVID-19, Viral Pneumonia, Normal) on the COVID-19 Radiography Database. Employing transfer learning with architectural modifications, including a tailored classification head and regularization techniques, the model achieves superior performance. Evaluated via accuracy, F1-score (macro-averaged), AUROC (macro-averaged), precision (macro-averaged), recall (macro-averaged), inference… More >

Manaliben Amin^*

Journal on Artificial Intelligence, Vol.8, pp. 1-18, 2026, DOI:10.32604/jai.2026.073895 - 07 January 2026

Abstract Artificial intelligence (AI) is steadily making its way into pharmaceutical validation, where it promises faster documentation, smarter testing strategies, and better handling of deviations. These gains are attractive, but in a regulated environment speed is never enough. Regulators want assurance that every system is reliable, that decisions are explainable, and that human accountability remains central. This paper sets out a Human-in-the-Loop (HITL) AI approach for Computer System Validation (CSV) and Computer Software Assurance (CSA). It relies on explainable AI (XAI) tools but keeps structured human review in place, so automation can be used without creating… More >

Zheng Yin, Zhao Zhang^*

Journal on Artificial Intelligence, Vol.7, pp. 633-645, 2025, DOI:10.32604/jai.2025.074450 - 24 December 2025

Abstract Power load forecasting load forecasting is a core task in power system scheduling, operation, and planning. To enhance forecasting performance, this paper proposes a dual-input deep learning model that integrates Convolutional Neural Networks, Gated Recurrent Units, and a self-attention mechanism. Based on standardized data cleaning and normalization, the method performs convolutional feature extraction and recurrent modeling on load and meteorological time series separately. The self-attention mechanism is then applied to assign weights to key time steps, after which the two feature streams are flattened and concatenated. Finally, a fully connected layer is used to generate More >

Chirag Devendrakumar Parikh^*

Journal on Artificial Intelligence, Vol.7, pp. 571-584, 2025, DOI:10.32604/jai.2025.073765 - 01 December 2025

Abstract With the increasing complexity and scale of hyperscale data centers, the requirement for intelligent, real-time power delivery has never been more critical to ensure uptime, energy efficiency, and sustainability. Those techniques are typically static, reactive (since CPU and workload scaling is applied to performance events that occur after a request has been submitted, and is thus can be classified as a reactive response.), and require manual operation, and cannot cope with the dynamic nature of the workloads, the distributed architectures as well as the non-uniform energy sources in today’s data centers. In this paper, we… More >

Miah A. Robinson, Abdulghani M. Abdulghani, Mokhles M. Abdulghani, Khalid H. Abed^*

Journal on Artificial Intelligence, Vol.7, pp. 615-632, 2025, DOI:10.32604/jai.2025.073535 - 01 December 2025

Abstract Ensuring the safety of Artificial Intelligence (AI) is essential for providing dependable services, especially in various sectors such as the military, education, healthcare, and automotive industries. A highly effective method to boost the precision and performance of an AI agent involves multi-configuration training, followed by thorough evaluation in a specific setting to gauge performance outcomes. This research thoroughly investigates the design of three AI agents, each configured with a different number of hidden units. The first agent is equipped with 128 hidden units, the second with 256, and the third with 512, all utilizing the… More >

Yu Tao¹, Tian Yin², Yang Jie^1,*

Journal on Artificial Intelligence, Vol.7, pp. 549-570, 2025, DOI:10.32604/jai.2025.072727 - 01 December 2025

Abstract To address the issue of insufficient accuracy in attitude estimation using Inertial Measurement Units (IMU), this paper proposes a multi-sensor fusion attitude estimation method based on an improved Error-State Kalman Filter (ESKF). Several adaptive mechanisms are introduced within the standard ESKF framework: first, the process noise covariance is dynamically adjusted based on gyroscope angular velocity to enhance the algorithm’s adaptability under both static and dynamic conditions; second, the Sage-Husa algorithm is employed to estimate the measurement noise covariance of the accelerometer and magnetometer in real-time, mitigating disturbances caused by external accelerations and magnetic fields. Additionally,… More >

Sewwandi Tennakoon¹, Eric Danso¹, Zhenjie Zhao^2,*

Journal on Artificial Intelligence, Vol.7, pp. 517-547, 2025, DOI:10.32604/jai.2025.072628 - 01 December 2025

Abstract Generative Artificial Intelligence (GenAI) systems have achieved remarkable capabilities across text, code, and image generation; however, their outputs remain prone to errors, hallucinations, and biases. Users often overtrust these outputs due to limited transparency, which can lead to misuse and decision errors. This study addresses the challenge of calibrating trust in GenAI through a human centered testing framework enhanced with adaptive explainability. We introduce a methodology that adjusts explanations dynamically according to user expertise, model output confidence, and contextual risk factors, providing guidance that is informative but not overwhelming. The framework was evaluated using outputs… More >

Tesfaye Adisu Tarekegn^1,*, Taye Girma Debelee^1,2

Journal on Artificial Intelligence, Vol.7, pp. 585-613, 2025, DOI:10.32604/jai.2025.067333 - 01 December 2025

Abstract The identification of defect types and their reduction values is the most crucial step in coffee grading. In Ethiopia, the current coffee defect investigation techniques rely on manual screening, which requires substantial human resources, time-consuming, and prone to errors. Recently, the deep learning driven object detection has shown promising results in coffee defect identification and grading tasks. In this study, we propose KN-YOLOv8, a modified You Only Look Once version-8 (YOLOv8) model optimized for real-time detection of coffee bean defects. This lightweight network incorporates effective feature fusion techniques to accurately detect and locate defects, even… More >

Meile Li¹, Yinuo Jiang^2,*

Journal on Artificial Intelligence, Vol.7, pp. 509-516, 2025, DOI:10.32604/jai.2025.073792 - 28 November 2025

Abstract This paper conducts topic mining and analysis of research literature in the domestic smart library field based on the BERTopic model, aiming to reveal its topic development context and evolution trends. Journal literature in the smart library field collected by CNKI (China National Knowledge Infrastructure) from 2015 to 2024 was analyzed using the BERTopic model and dynamic topic modeling for topic mining and evolution trend analysis. The study found that the domestic smart library field involves multiple core topics, identifying a diversified topic structure centered around “data”, “user”, “5g”, etc. The research results provide data More >

Palak Bari, Gurnur Bedi, Khushi Joshi, Anupama Jawale^*

Journal on Artificial Intelligence, Vol.7, pp. 499-508, 2025, DOI:10.32604/jai.2025.072531 - 17 November 2025

Abstract This study investigates sarcasm detection in text using a dataset of 8095 sentences compiled from MUStARD and HuggingFace repositories, balanced across sarcastic and non-sarcastic classes. A sequential baseline model (LSTM) is compared with transformer-based models (RoBERTa and XLNet), integrated with attention mechanisms. Transformers were chosen for their proven ability to capture long-range contextual dependencies, whereas LSTM serves as a traditional benchmark for sequential modeling. Experimental results show that RoBERTa achieves 0.87 accuracy, XLNet 0.83, and LSTM 0.52. These findings confirm that transformer architectures significantly outperform recurrent models in sarcasm detection. Future work will incorporate multimodal More >