Transactions on Machine Intelligence2821-16936420231001An Improved Object Tracking Technique for Remote Weapon Station Using Yolov5_Deepsort_Dlib Architecture18219718143810.47176/TMI.2023.182ENO. EzekielOlorunsholaComputer Science Department,
Air Force Institute of Technology,
Kaduna State,
NigeriaM. EkataIrhebhudeComputer Science Department,
Nigerian Defence Academy,
NigeriaA. EseogheneEvwiekpaefeComputer Science Department, Nigerian Defence Academy, Kaduna, NigeriaJournal Article20230902This paper introduces an advanced tracking object architecture named DeepSORT_YOLOv5_Dlib. Building upon the DeepSORT_YOLOv3 framework, the study [1] integrates the Digital Library's correlation tracker into the traditional DeepSORT_YOLOv3 to minimize identity switches. Notably, the architecture is designed to operate in parallel, enhancing its operational speed. Experimental results indicate that the proposed approach outperforms the conventional DeepSort_YOLOv3, showcasing reduced identity switches and increased operational speed across various video testing scenarios. The custom model employed in this study adopts a confidence threshold of 0.2 and an image size of 416 x 416, consistent with the training size. To boost detection within YOLOv5, the model incorporates the Slicing Aided Hyper Inference (SAHI) technique. The overall inference speed in this study reaches 314.8fps, a notable improvement compared to Dang's 218.6fps. Evaluation using the COCO dataset demonstrates the model's precision at 0.98 and a recall of 0.81. Additionally, the proposed custom model exhibits a MOTA of 0.86, surpassing the benchmark's 0.83. Notably, our model achieves a significantly lower identity switch count of 1881 compared to the benchmark's count of 2288. Furthermore, it outperforms the benchmark in object detection capabilities. By incorporating SAHI inference with YOLOv5, the study enhances detection accuracy, resulting in an overall tracking accuracy improvement from 56% to 79%. These findings highlight the efficacy of the proposed custom model in achieving superior performance in object tracking and detection.https://www.tmachineintelligence.ir/article_181438_3751a70b63e8bb910830a9db97f80521.pdfTransactions on Machine Intelligence2821-16936420231001Diagnosis of Atherosclerosis of the Coronary Arteries of The Heart with Data Mining and Machine Learning Techniques19820717138610.47176/TMI.2023.198ENS.AsadzadehDepartment of Computer Engineering, Islamic Azad University, Shiraz Branch, Shiraz, Iran0000-0002-2487-1631M. A.ShayeganAssistant Professor, Department of Computer Engineering, Islamic Azad University, Shiraz Branch, Shiraz, Iran0000-0001-7394-4772Journal Article20230902Cardiovascular diseases are the leading cause of death in the world. In this regard, the rapid and timely diagnosis of heart diseases and the prediction of certain risk events associated with the cardiovascular system are among the top priorities of researchers. Due to the risks of invasive diagnostic methods in coronary artery disease, such as angiography, providing a suitable and non-invasive method for timely diagnosis, increasing accuracy, reducing errors in decision-making, reducing treatment costs and improving the quality of services provided by physicians has been the main goal of this research. In the implementation of this practical research, the Cleveland medical data set, consisting of 270 samples with 76 features, and Z-AlizadehSani data set, consisting of 303 samples with 54 features, available in the UCI standard data repository, were used. Initially, preprocessing and feature selection, followed by modeling, data processing and analysis was performed by examining the effect of disease parameters on coronary artery stiffness using a combination of machine learning algorithms. The proposed system, based on accuracy, sensitivity, specificity, and AUC indices, was able to achieve the best performance with the lowest error compared to similar research. Based on the results obtained, the proposed model can prevent potential adverse effects and damages of some invasive procedures such as angiography in patients who do not need it. Moreover, the system can help physicians triage patients who definitely need these diagnostic procedures in order to receive timely treatment with the highest precision.https://www.tmachineintelligence.ir/article_171386_8d4967a38d65e1d1cd2251cfe4d00012.pdfTransactions on Machine Intelligence2821-16936420231001Solving Urban Routing Problem in Supply Chain by Optimizing PSO Algorithm20822017357010.47176/TMI.2023.208ENZ.SaadiDepartment of Computer Engineering, Malayer Branch, Islamic Azad University, Malayer, IranJournal Article20230902Addressing the effective distribution of service requests among vehicles in the supply chain stands out as a key hurdle in supply chain management, commonly referred to as vehicle routing with traffic balancing. The optimized vehicle routing, coupled with traffic balancing strategies, emerges as a pivotal factor contributing to heightened customer satisfaction, reduced delivery times, enhanced vehicle utilization, diminished service request demands, and an overall elevation in service quality within the supply chain. To address this, a proposed method involves real-time assessment of service requests and vehicle conditions, enabling balanced routing based on the current operational context. Given the NP-Hard complexity associated with the vehicle routing problem involving traffic balancing in the supply chain, leveraging optimization algorithms, such as PSO, proves to be a more efficient approach. This study introduces a PSO optimization algorithm tailored for the aforementioned challenge. By integrating real-time conditions of service requests and vehicles within the supply chain, the algorithm strategically selects optimal routes for each vehicle and service request. The PSO optimization algorithm undergoes simulation in Python software, undergoes evaluation, and is analyzed alongside comparable routing methods. The assessment outcomes reveal a reduction in distance traveled and total delivery time achieved through the application of the PSO optimization algorithm.https://www.tmachineintelligence.ir/article_173570_ebbfe3ec1c5b34e3750828e87a25aeae.pdfTransactions on Machine Intelligence2821-16936420231001Design Of a New Optimal Controller for a Particular Class of Chaotic Systems Using the Artificial Bee Colony Algorithm (ABC)22123518144210.47176/TMI.2023.221ENM.Khoshhal RudposhtiDepartment of Electrical Engineering, Langarud Branch, Islamic Azad University, Langarud, Iran0000-0001-8399-3930Journal Article20230511The primary aim of this paper is to devise an optimal regulator for stabilizing a distinct class of chaotic systems through a systematic two-step approach. Initially, the chaotic system undergoes transformation into state-dependent equations. Subsequently, the State Dependent Riccati Equation (SDRE) is tackled via the power series method, facilitating the determination of the optimal control law. Ensuring a suitable regulatory response involves the utilization of an intuitive optimization algorithm of a naturalistic nature, with a focus on optimizing the weight matrices within the SDRE equation. Employing the Artificial Bee Colony (ABC) algorithm, we derive the weighted matrices, leveraging the honey bee algorithm to fine-tune the gain coefficients by minimizing the chosen fitness function. The fitness function, represented as the sum of squares of system state errors, proves instrumental in achieving effective stabilization of the chaotic system, minimizing error, enhancing response speed, and reducing control costs. Through simulation, we scrutinize the effectiveness of regulators designed to stabilize and control chaotic systems, particularly comparing the regulatory performance of this algorithm against the SDRE method.https://www.tmachineintelligence.ir/article_181442_08b18b003542681939b50381abb2096d.pdfTransactions on Machine Intelligence2821-16936420231001Optimal Routing-Clustering Aware of Energy Consumption in Wireless Sensor Networks based on Deep Tree Learning23624718633110.47176/TMI.2023.236ENB.SalehDepartment of Information Technology, Sabzevar Branch, Islamic Azad University, Sabzevar, Iran0000-0002-3318-397XA. A.NeghabiPh.D., Department of Computer Engineering, Sabzevar Branch, Islamic Azad University, Sabzevar, Iran0000-0003-2147-7441Journal Article20230602Presently, the application of Wireless Sensor Networks (WSNs) poses challenges across various domains, with the most prominent being the energy consumption of sensor batteries. Sensor nodes, dispersed in diverse geographical environments for their designated purposes, rely on batteries for data collection. The deployment of sensor nodes induces energy losses during data collection and transmission, particularly in routing data, which demands substantial energy. To tackle this issue, clustering is employed either before or concurrently with routing. This article explores the implementation of clustering-routing alongside sleep and wake scheduling in sensor nodes to effectively conserve energy. The study introduces the optimal OCADR (Constrained Anisotropic Diffusion Routing) protocol, enhancing it with the DAVL (Deep Adelson-Velskii and Landis) tree rotation clustering algorithm. The research reveals that this innovative approach offers improved scheduling in terms of sensor nodes' sleep and wake time compared to prior methods. Moreover, it efficiently transmits packets to the base station through the head clusters. The initial energy allocation was 50 Joules, and after simulation using this method, only 22 Joules were consumed, leaving 28 Joules for network survival an advancement surpassing earlier methodologies.https://www.tmachineintelligence.ir/article_186331_35ee912d84159229f36e445ceab59063.pdfTransactions on Machine Intelligence2821-16936420231001Economic Load Spreading by Considering the Presence of Electric Vehicles in Order to Reduce Pollution Particles and To Consider Demand Response24826518633410.47176/TMI.2023.248ENF.BadakhshaniMaster's student, Electrical Department, Abadeh Branch, Islamic Azad University, Abadeh, IranN.GoudarziAssistant Professor, Department of Nuclear Engineering, Abadeh Branch, Islamic Azad University, Abadeh, Iran0009-0003-8733-1856Journal Article20230510The development of power networks has expanded the production and consumption of electrical energy into highly competitive areas, including the use of renewable energy. Solar energy, for example, not only reduces fuel consumption but also lowers the long-term cost of electricity production by requiring only initial capital investment. This research focuses on economic dispatching in the network. The aim of this research is to decrease environmental pollution and energy production costs. The subject being investigated exhibits non-linear properties in many areas, and cannot be viewed as a linear problem. Various tools are utilized in this research to achieve the main objective. One tool utilized in this research is the use of solar production sources and energy storage elements to shift part of the load during peak hours to low-load hours. Electric cars are used as an example. The particle swarm algorithm is employed to optimize the presence hours of electric vehicles and the amount that each power plant unit should produce. However, this research has implemented restrictions for the optimal use of these units. These restrictions will be explained in detail below.https://www.tmachineintelligence.ir/article_186334_11aa2fb7a9ca266b6705bd2a2c41818f.pdf