Transactions on Machine Intelligence2821-16937420241201A Review of Different Approaches on Determination of Convergence Control Parameter in Homotopy Analysis Method23824520511810.47176/TMI.2024.238ENM.JaliliDepartment of Mathematics, Neyshabur Branch, Islamic Azad University, Neyshabur, IranJournal Article20240630One of the principal components of the Homotopy Analysis Method (HAM) is the determination of the convergence control parameter, which plays a pivotal role in ensuring the accuracy and efficiency of solutions obtained using HAM. The convergence control parameter directly impacts the rate of convergence and the precision of the method, making its proper determination essential for solving nonlinear problems. This study aims to systematically compare the performance of several approaches for determining the convergence control parameter in HAM. By examining different methods, the paper highlights their respective strengths, weaknesses, and applicability to a range of nonlinear problems. Numerical experiments and theoretical analysis are conducted to assess the accuracy and convergence rate associated with each approach. Particular attention is given to identifying strategies that achieve a balance between computational efficiency and solution precision. The results provide valuable insights into the impact of the convergence control parameter on HAM's performance and offer guidelines for selecting the most suitable approach for various types of problems. This study contributes to advancing the application of HAM in solving nonlinear equations, enhancing its utility in scientific and engineering contexts.https://www.tmachineintelligence.ir/article_205118_f0b734ecbce2e00a2bba209f13e317c8.pdfTransactions on Machine Intelligence2821-16937420241201Breast Cancer Histopathology Image Classification Using a Set of Deep Learning Models and VGG16 and VGG19 Architectures24625620511910.47176/TMI.2024.246ENM.Shirin ManeshMaster of Science in Biomedical Engineering, Islamic Azad University, Garmsar Branch, Iran.V.RazmavarAssistant Professor, Department of Biomedical Engineering, Faculty of Engineering, Islamic Azad University, Garmsar Branch, Iran.Journal Article20240606Breast cancer is one of the most prevalent and serious public health challenges worldwide, being the leading cause of cancer-related deaths among women. Early detection is a critical factor in improving survival rates, as it allows for timely intervention and treatment. The complexity of diagnosing breast cancer from histopathology images has led to the development of advanced techniques using artificial intelligence (AI) and machine learning. This study introduces a novel deep learning ensemble approach to classify breast cancer histopathology images using publicly available datasets. The primary objective of this research is to improve the classification accuracy of breast cancer images by leveraging a combination of two deep learning models. The proposed approach utilizes the VGG16 and VGG19 models, which were both fine-tuned to enhance their performance. The results demonstrate that the ensemble method, which averages the predicted probabilities from both models, leads to a more robust classifier. Specifically, the fine-tuning of the VGG16 and VGG19 models contributes significantly to improving the model’s performance. The ensemble model exhibits competitive results in classifying complex histopathology images of breast cancer, with a recall value of 97.73% for the cancer class in both the full training and fine-tuning approaches. This research highlights the effectiveness of ensemble learning in medical image classification, paving the way for more accurate and reliable tools in the diagnosis of breast cancer.https://www.tmachineintelligence.ir/article_205119_94f5389b695a22d83b138789cff1c060.pdfTransactions on Machine Intelligence2821-16937420241201Video Tracking Using Correlation Measurement25726820512010.47176/TMI.2024.257ENA.ToobaeiDepartment of Electrical and Telecommunications Engineering, Faculty of Electrical and Computer Engineering, University of Sistan and Baluchestan, Zahedan, IranF.MehnaAssistant Professor, Department of Electrical and Telecommunications Engineering, University of Sistan and Baluchestan, Zahedan, IranJournal Article20240406Video tracking is one of the fundamental and widely used topics in the field of computer vision, with applications ranging from surveillance and security systems to autonomous vehicles and human-computer interaction. Despite extensive research and significant advancements in this area, numerous challenges continue to affect the performance and reliability of visual tracking systems. Among these challenges are occlusion, where objects are partially or completely hidden from view; lighting changes, which can alter the appearance of objects; scale and size variations; rapid object movements; background clutter; and computational complexity, which hinders real-time processing capabilities. In this paper, a particle filter is employed as the motion model to predict and estimate the states of the target object over time. The observation model evaluates the likelihood of the predicted states based on their correlation with a predefined pattern or template. By leveraging correlation measurement for state evaluation, the proposed video tracking method can address several of the common challenges, such as changes in lighting and partial occlusion. Moreover, because correlation measurement typically requires lower computational resources compared to more complex tracking algorithms like deep learning-based methods, the proposed approach enables faster processing speeds. Consequently, it offers improved performance for real-time applications where both accuracy and efficiency are critical.https://www.tmachineintelligence.ir/article_205120_a96e4e0c908ef37bb2520b3e661a325f.pdfTransactions on Machine Intelligence2821-16937420241201Brain Tumor Detection in MRI Images Using ResNet18 Convolutional Neural Network and Transfer Learning26927520512110.47176/TMI.2024.269ENM. M.Haj Hashem KhaniDepartment of Electrical Engineering (Electronics), South Tehran Branch, Islamic Azad University, Tehran, IranF.Maleki NodehiDepartment of Electrical Engineering (Electronics), South Tehran Branch, Islamic Azad University, Tehran, IranJournal Article20240713In this study, a deep learning-based brain tumor detection model is proposed using a Convolutional Neural Network (CNN) architecture, specifically the ResNet18 model. The aim is to develop an automated and accurate system capable of detecting brain tumors from MRI images, classifying them into two categories: “tumor present” and “no tumor.” To enhance performance and reduce the need for large-scale annotated medical datasets, the model employs transfer learning by initializing with pre-trained weights from the ImageNet dataset. The final fully connected layers of the ResNet18 network are fine-tuned to adapt to the specific binary classification task. The MRI dataset is divided into training and test sets, and preprocessing steps such as image resizing and normalization are applied to standardize inputs. After training for ten epochs, the model achieved promising results, including an accuracy of 84.31%, a precision of 79.31%, a recall of 92.00%, and an F1 score of 85.19%. These metrics indicate the model’s robustness in detecting tumors with high sensitivity and specificity. The experimental results suggest that the proposed method can effectively extract and interpret critical features from MRI scans, offering a reliable tool for assisting radiologists in early diagnosis and reducing the risk of human error in clinical decision-making.https://www.tmachineintelligence.ir/article_205121_86fe9cf2ee0ddd669a4565465586de21.pdfTransactions on Machine Intelligence2821-16937420241201Strategic Enhancement of Airline Maintenance Operations A KPI-Driven Approach for the Chief Line Maintenance Officer27628521399710.47176/TMI.2024.276ENS.A.MoghadasnianTarbiat Modares University, Tehran, IranN.SarviIslamic Azad University،Science and Research Branch, Tehran, IranJournal Article20240902This study investigates the pivotal role of Key Performance Indicators (KPIs) in strengthening and transforming airline maintenance operations, with particular emphasis on the strategic functions and leadership responsibilities of the Chief Line Maintenance Officer (CLMO). Employing a mixed-methods research design, the paper systematically examines critical dimensions including operational efficiency, regulatory and safety compliance, maintenance quality, financial and cost performance, employee engagement, workforce productivity, and the integration of advanced technologies such as digital monitoring and predictive maintenance tools. The results highlight how a well-structured, KPI-driven management framework can significantly enhance not only the efficiency and safety of airline maintenance processes but also the overall cost-effectiveness, decision-making, and long-term sustainability of operations. Moreover, the study sheds light on the organizational and cultural factors that affect the successful adoption of KPI-based strategies, underscoring the importance of leadership commitment, cross-functional collaboration, continuous improvement, and data-driven decision-making. By providing a detailed and adaptable framework, this paper offers valuable insights for both academic researchers and industry practitioners seeking to optimize aviation maintenance practices in an increasingly competitive and technologically complex environment. The findings contribute to the broader discourse on performance management and strategic leadership in the aviation sector, paving the way for future research and practical advancements.https://www.tmachineintelligence.ir/article_213997_1927f55bd606ebe61cf9536996b070e1.pdfTransactions on Machine Intelligence2821-16937420241201A Novel Frequency Compensation Scheme for Stabilizing Three-Stage Amplifiers for a Wide range of Load Capacitors28629521399810.47176/TMI.2024.286ENY.BelghisAzarDepartment of Electrical Engineering, Azarbaijan Shahid Madani University, Tabriz, IranJournal Article20240802This paper introduces an innovative frequency compensation method designed to enhance the stability of three-stage amplifiers without imposing limitations on capacitive loads (CL). Traditional multistage amplifiers often encounter stability challenges due to varying CLs, but the proposed scheme overcomes this issue through an advanced compensation strategy. The approach integrates Miller capacitors in series with embedded current buffers and incorporates a feedback network connecting the second stage to the first. By effectively tuning the quality factor of non-dominant poles, the method achieves robust stability across a broad range of capacitive loads. The design was implemented using standard 180 nm CMOS technology, operating at a nominal supply voltage of 1.8 V. The compact layout occupies only 0.0026 mm² and consumes 23.38 µA of current. Open-loop frequency response simulations across different CL values, ranging from 0 to 100 nF, demonstrated a high gain of 119.3 dB. The average unity-gain bandwidths were measured at 4.872 MHz, 2.4 MHz, and 90.11 kHz for no load, 0.1 nF, and 100 nF loads, respectively. The paper also evaluates the amplifier's output settling performance under varying load conditions. Configured as a buffer with a 0.4 V input step, the amplifier exhibited stable behavior across all tested CLs. The measured 1% settling times were 0.48 µs for a 100 pF load and 19.92 µs for a 100 nF load, highlighting the effectiveness of the proposed compensation scheme in ensuring stability and performance for three-stage amplifiers.https://www.tmachineintelligence.ir/article_213998_ec016997c5191b953ececd3f7fc1a51d.pdf