Applied and Computational Engineering

- The Open Access Proceedings Series for Conferences

Volume Info.

  • Title

    Proceedings of the 2023 International Conference on Machine Learning and Automation

    Conference Date

    2023-10-18

    Website

    https://2023.confmla.org/

    Notes

     

    ISBN

    978-1-83558-307-4 (Print)

    978-1-83558-308-1 (Online)

    Published Date

    2024-02-22

    Editors

    Mustafa İSTANBULLU, Cukurova University

Articles

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230702

    Control strategies modeling for robotic exoskeletons facilitating sit-to-stand transitions in geriatric and lower limb impaired

    With the growing global aging population, there's been an amplified societal emphasis on preserving the health of the elderly and enhancing their quality of life. In this scenario, robotic exoskeletons have emerged as a cutting-edge solution to assist the elderly and those with lower limb muscle deficiencies in Sit-to-Stand (STS) exercises. These exoskeletons adopt two main approaches: full assistance for those with entirely weakened lower limbs and partial assistance for those with some remaining muscle strength. This article introduces two modeling methods and concepts for these control strategies, aligning with the full and partial assistance directions, respectively. Both approaches hinge on the Lagrange equation as their foundational structure, integrating distinct kinematic designs to form their individual dynamic models. Based on this, the models are further adapted to address the specific risks associated with STS activities as per each strategy. Research outcomes highlight that by assessing the wearer's EMG signal, the partial assistance strategy considerably mitigates the lower limb muscle strength required for STS under conditions such as low-speed, medium-speed, sitback-like, and step-like. This not only improves balance but also augments the likelihood of successful STS execution, consequently diminishing fall incidents.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230703

    Research on grasping model based on visual recognition robot arm

    This article mainly systematically describes the research based on the visual recognition robotic arm. With the advancement of science and technology, the robot industry has also seen significant improvement in recent years. The amount of the use of robots, especially robotic arms, is increasing rapidly. After large-scale improvements, some companies have abandoned simple traditional robotic arms that have been eliminated from the industry and cannot meet the demands of the industry but install more high-tech elements on the robotic arm for use. In the upgrade of the robot arm, whether it is for the system or hardware, or software, there are some breakthrough improvements. Some companies use visual sensors in robotic arms to find and detect target objects and perform actions. Due to the gradual improvement of visual recognition technology, visual recognition technology has been widely used. Based on the understanding of the field of the visual recognition robot arm and consulting a lot of literature, this paper summarizes the current situation of the existing visual recognition robot arm and analyzes the principle and design of the visual recognition grasping robot arm. This paper focuses on analyzing how the existing visual recognition analysis works, how the robot arm recognizes the coordinates of the object and analyzes the object, and then grabs the object and puts it into the corresponding position, to achieve flexible and smooth use, then put it into the industry. After understanding the current situation, this paper will discuss and analyze the existing CNN model and transformer model for visual recognition applications, analyze and explain the principles and characteristic analysis methods of these two models, while comparing the two models, analyze the advantages and disadvantages, and propose areas that can be optimized.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230706

    Advancements in VLSI low-power design: Strategies and optimization techniques

    As production technology advances, integrated circuits are increasing in size, leading to a corresponding rise in power consumption if not properly optimized. Consequently, the optimization of integrated circuit power consumption has gained paramount significance. This paper provides an overview of the theoretical and research developments in Very Large Scale Integration (VLSI) low-power design. Initially, the paper delves into the components of VLSI power consumption, elucidating the origins of various power consumption types and the factors influencing their magnitude. Subsequently, existing power reduction technologies are examined, including transistor-level optimization, gate-level optimization, and system-level power optimization. The principles, applicable power consumption types, as well as their respective advantages and drawbacks are analysed. The paper also introduces methods for evaluating VLSI power consumption and summarizes the characteristics, advantages, and disadvantages of high-level power estimation and low-level power estimation. Ultimately, it underscores the importance of considering multiple power optimization strategies during VLSI design and discusses research approaches for achieving low power consumption. This comprehensive exploration contributes to the enhancement and optimization of VLSI design efforts.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230707

    Design and optimization of CMOS based 4-bit comparator

    The paper primarily focuses on optimizing circuit delay and energy consumption, specifically in gate-level circuits. Three methods are employed for circuit optimization. The first method aims to minimize transistor usage to reduce both delay and energy consumption. The second method involves prioritizing logic gates based on underlying hardware, favoring simpler circuit structures whenever possible, given that our design primarily revolves around logic gates. The third method entails adjusting the number of stages to enhance delay optimization. To validate these rules, three distinct circuits were designed to implement a 4-bit absolute value comparator, each corresponding to one of the rules. Through simulation, calculation, and comparison, the best circuit was identified, providing validation for the rules. The second part of the paper shifts its focus towards optimizing delay and energy consumption by adjusting the sizing of logic gates and the supply voltage to achieve optimal overall performance. In conclusion, further research is needed to corroborate these three rules and identify additional rules, laying the foundation for intelligent circuit optimization.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230708

    Application and analysis of face matching based on the Siamese model in face recognition

    In recent years, face recognition technology, and face matching particularly have broadened the application fields in various aspects of society. It is considered a combination of deep learning architecture and face recognition technology, which has been used for personal information security and safety efficiently for many years. For this, this paper aims to investigate the practical method of utilizing Siamese models to enhance the accuracy and efficiency of face matching systems. The existing challenges of low accuracy and slow recognition rates in face matching applications have been approved to be solvable by utilizing the capabilities of the Siamese model. Experimental analysis and comments from relevant practitioners demonstrate the effectiveness and potential of the Siamese model in enhancing the performance of face matching systems. To conclude, the Siamese model is introduced as a robust and efficient tool in the field of face recognition. It provides higher accuracy and efficiency compared to the traditional feature-based models. Its adaptability and advancements bring the potential to revolutionize face-matching applications and overcome current limitations. The findings from the experiments demonstrate that the utilization of the joint model can significantly enhance the performance of the matching system. The proposed model offers a potential solution to address the issue of low accuracy during the face matching phase.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230709

    Genetic protein sequence analysis based on sequence alignment techniques for time series data

    This study aims to explore the application and effectiveness of sequence comparison techniques in dealing with missing and outliers in time series data. First, the data are pre-processed by convolutional neural network (CNN) and recurrent neural networks (RNN) to remove noise and outliers. Then, time series data at different time points are compared and analysed using the comparison loss function to identify changes and differences in the data. Finally, the prediction performance of different models is evaluated using a variety of assessment metrics, and the results are compared and analysed to verify the effectiveness of the sequence comparison technique in dealing with missing and outliers. The experimental results show that the sequence comparison technique can effectively deal with missing and outliers in time series data, providing important insights for further research on the application and development of the sequence comparison technique. Future research can explore the application of sequence comparison techniques in more fields to optimize model performance and improve accuracy and stability.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230710

    Predictive modelling based on statistical modeling of logistic regression for heart disease

    The heart is the core driving force for the continuation of human life, and the disease of this organ is bound to be fatal. There are two main types of heart disease. Congenital diseases are caused by developmental problems in unborn children. These problems, mainly in the heart, can damage various parts of the heart. Acquired sexually transmitted diseases are diseases caused by environmental factors and their own growth and development after birth. The purpose of the project model is to predict heart disease and analyse the main types of heart disease in the population. In the whole research process, the most important thing is the establishment of the model. The algorithm principle of this model is logistic regression. Logistic regression is used to make predictions and probability calculations on the data. Through such algorithms, modelling techniques can be used to predict the impact of pathogenic factors on the probability of heart disease. In addition, prevention of heart disease can be improved with accurate and convenient model predictions that can be tailored to the population that fits the predictions. This method can improve the technical level and treatment level of the hospital, and can also reduce the harm caused by heart disease.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230711

    The prediction and analysis of heart disease using XGBoost algorithm

    Heart diseases remain a global health concern, with their intricate aetiology and multifactorial risk factors making early diagnosis challenging. Recognizing the pressing need for accurate prediction tools, this research ventured into harnessing the power of machine learning, notably the Xtreme Gradient Boosting (XGBoost) algorithm, to fill this gap. The main object is to devise a robust predictive framework capable of early and accurate identification of heart disease. Specifically, our methodology unfolded systematically, beginning with data preprocessing, then delving into incisive feature selection, rigorous model training, and finally, thorough evaluation. This study is meticulously conducted on the ‘heart.csv’ dataset, a comprehensive repository of cardiovascular data points. The experimental outcomes were nothing short of revelatory. Not only did the XGBoost model manifest superior performance metrics, but its precision also outpaced several contemporary models referenced in existing literature. Ultimately, our findings underscore the profound potential of the XGBoost algorithm in heart disease predictions. Beyond academic intrigue, this research holds tangible implications for healthcare practitioners, potentially offering a novel tool for early interventions and patient management.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230712

    Prediction of stress levels in sleep patterns based on random forest

    The prevalence of stress in contemporary society has emerged as a significant concern, exerting a profound influence on our daily lives. The objective of this study is to predict stress levels in sleep patterns through the utilization of a machine learning algorithm known as random forest. The significance of stress detection has increased due to its potential to induce various issues such as insomnia and depression. The examination of stress can assist individuals in mitigating the adverse effects associated with prolonged exposure to stress. The study commences with the preprocessing phase, followed by an exploratory data analysis, subsequent dataset splitting, identification of significant features, and concludes with model training. The utilization of the random forest model can enhance the comprehension of the association between sleeping characteristics and levels of stress. Furthermore, it produces a f1-score of 98 percent, indicating a strong predictive capability for determining stress levels in sleep patterns. The proposed method can effectively predict stress levels during sleep mode. This study can provide an effective model for society to prevent people's psychological problems in advance.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230713

    Applications and challenges of GAN in AI-powered artistry

    In the evolving landscape of artificial intelligence (AI), Generative Adversarial Network (GAN), introduced in 2014 by Goodfellow and team, has emerged as a vital pillar in deep learning. Designed around the concept of adversarial learning, GAN consists of a generator and a discriminator working in tandem, with the former creating counterfeit data samples and the latter distinguishing between genuine and counterfeit ones. The paper delves deep into GAN’s underlying architecture, its modified variants like DCGAN, WGAN, WGAN-GP, and CGAN, and its expansive applications in the realm of AI-powered artistry. Notably, applications like Stable Diffusion and NovelAI have demonstrated GAN’s proficiency in crafting visually stunning and diverse artistic outputs. However, this evolution isn’t without challenges. The ambiguities surrounding copyright ownership of AI-generated art and the potential disruption of the traditional art sector raise critical questions. As AI continues to redefine the boundaries of art, it’s imperative to ensure its responsible and beneficial integration into society.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230714

    Improved small-object detection using YOLOv8: A comparative study

    In the last decade or so, deep neural networks have evolved at a rapid pace, where computer vision has been constantly refreshing its best performance and has been integrated into our lives. In the field of target detection, YOLO model is a popular real-time target detection algorithm model that is fast, efficient, and accurate. This research aims to optimize the latest YOLOv8 model to improve its detection of small objects and compare it with another different version of YOLO models. To achieve this goal, we used the classical deep learning algorithm YOLOv8 as a benchmark and made several improvements and optimizations. We optimized the definition of the detection head, narrowed its perceptual field, and increased its number, allowing the model to better focus on the detailed information of small objects. We compared the optimized YOLOv8 model with other classical YOLO models, including YOLOv3 and YOLOv5n. The experimental results show that our optimized model improves small object detection with higher accuracy. This research provides an effective solution for small object detection with good application prospects. With the continuous development and improvement of the technology, we believe that the YOLO algorithm will continue to play an essential role in object detection and provide a reliable solution for various real-time applications.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230715

    Self-avatar: Monocular 3D human reconstruction from RGB image

    3D human position and shape estimate are crucial in many computer vision applications. Despite the fact that there are numerous deep learning techniques designed to handle this problem, they frequently only use training networks with RGB images from a single point of view. In this paper, a unique approach to solve this issue is proposed by combining a regression-based multi-view picture learning loop with an optimization-based multi-view model. This is because some public datasets are collected by multi-view camera systems. A parameterized human body model's position and shape parameters are initially deduced by a convolutional neural network (CNN) from multi-view photos. This work then introduces an enhanced multi-view optimization method called MV-SMPLify, which aligns the SMPL model with multi-view images by using the regressed pose and shape as beginning values. Following that, the CNN model's training can be monitored using the optimum parameters. The Self-avatar project as a whole is a self-supervised framework that combines the advantages of both the CNN method and the optimization-based strategy. Additionally, the use of multi-view photos improves thorough supervision during training. This methodology outperforms earlier methods in a variety of ways, according to qualitative and quantitative testing using open datasets.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230716

    Path planning algorithms of sweeping robots

    Different categories of path planning algorithms for sweeping robot are introduced, including Dijkstra algorithm and A*Algorithm in Traditional path-planning Algorithm, PRM Algorithm and RRT Algorithm in sampling algorithm, and Ant Colony Optimization Algorithms and Genetic algorithms in Intelligent bionic algorithm. Each algorithm has its principles and features introduced. At the same time, several algorithms are compared, and summarized, each algorithm has its advantages and disadvantages, in the future development should be combined with their strengths to optimize the path planning algorithm of the sweeping robot.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230718

    Exploring data storage innovations: From DNA to holography

    This paper delves into the evolution of data storage technologies, focusing on the solid-state drive (SSD) and hard disk drive (HDD) and their working principles. It explores the advancements in DNA as a storage medium, highlighting its exceptional data density and potential as a long-term storage solution. The paper discusses the challenges and limitations of DNA data storage, emphasizing the need for further research to overcome these hurdles. Additionally, the paper introduces holographic data storage as a novel optical storage technology and discusses its potential applications. It also touches on the decline of the optical disc industry due to the emergence of alternative storage technologies. The use of metal nanoparticles (MNPs) in optical disks is explored, emphasizing their role in high-contrast imaging and data storage. The paper concludes by highlighting the diverse methods of data storage, each with its unique advantages and drawbacks, and anticipates future improvements in electronic information processing technology, leading to faster data transmission and increased data storage capacity within the same volume of storage media.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230721

    Battery-aware federated learning: Challenges and solutions

    Smartphone battery life is a pivotal factor in consumers' purchasing decisions. Recent years have witnessed a surge in studies focusing on smartphone energy management, with data-driven energy management systems offering solutions to prolong battery life. Federated Learning (FL) emerges as a promising distributed learning algorithm, enabling wireless devices to upload locally trained models, fostering collaborative learning without exposing sensitive data. This paper explores the FL process, particularly the Federated Averaging (FedAvg) approach, which excels in scenarios with homogeneous data. In the era of burgeoning data generation, traditional cloud computing systems face limitations, driving the adoption of Edge Computing (EC), which processes data closer to its source, enhancing response times. To make FL efficient for e-commerce, resource constraints must be addressed. This involves techniques like local updates and model compression, which reduce communication overhead. However, FL brings challenges related to data distribution heterogeneity and privacy concerns. Solutions like differential privacy, encryption, and access control are discussed. In conclusion, this paper presents an overview of smartphone battery life, data-driven energy management, and the potential of FL, emphasizing its relevance in the age of EC. By addressing resource limitations and privacy issues, FL holds promise for efficient data processing.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230725

    Intelligent vehicle navigation systems and autonomous driving technology: A comprehensive analysis

    This paper conducts a comprehensive study and analysis of intelligent vehicle navigation systems and autonomous driving technology. We review the historical development of autonomous driving technology, discuss key concepts such as perception, decision-making, and control, explore various types of autonomous vehicles, and examine various aspects of intelligent vehicle navigation systems. Additionally, we investigate safety, reliability, legal frameworks in the field of autonomous driving, as well as future trends and ethical considerations. Finally, we summarize the main findings of the research and provide recommendations for future studies.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230729

    Investigation related to detection of Intracranial Hemorrhage based on edge impulse enhanced CT scanning

    Intracranial Hemorrhage (ICH) is a critical medical condition demanding rapid and precise diagnosis, typically achieved through Computerized Tomography (CT) scans. This research investigates the potential of the Edge Impulse platform, a symbol of progress in edge computing, for the automatic detection of Intracranial Hemorrhage (ICH). The study leverages RGB images extracted from CT scans, employing transfer learning techniques. By utilizing the “brain ct hemorrhage AMINE dataset” available on Kaggle, this research combines Convolutional Neural Networks (CNNs) with the efficiency and adaptability offered by the MobileNet framework in a novel approach to address this diagnostic challenge. To ensure the models strength, robustness, applicability and a useful approach has been used, this study tested setups of the neural network to find the most effective ones. These setups involved changing parameters like resolution (ρ) and width multipliers (α) which greatly impact the model’s diagnostic performance. The remarkable result was observed in a configuration, with a resolution of 160x160 pixels and a width multiplier of 0.5. After optimization this specific setup achieved an outstanding diagnostic accuracy rate of 99.8% with negligible loss. This accomplishment highlights how edge computing, through Edge Impulse can significantly improve and speed up ICH diagnostic procedures.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230730

    Machine learning models for COVID-19 diagnosis based on medical images and audio

    The COVID-19 pandemic presents a significant danger to human health, with far-reaching consequences for the global economy and political dynamics. It presents complex challenges in terms of rapid and accurate diagnosis, where machine learning holds potential to enhance diagnostic speed and precision while reducing time and resource burdens. Consequently, this research employs CT images and cough audio recordings as training data to create machine learning models for data classification, with the goal of aiding in the diagnosis of COVID-19. Using the Edge Impulse platform, a Convolutional Neural Network, MobileNetV2, is customized for efficient image recognition. On the audio front, the preprocessing phase encompassed three distinct feature extraction techniques: Mel Frequency Cepstral Coefficients, Mel-Filterbank Energy (MFE), and Mel spectrogram. Subsequently, model frameworks were meticulously adjusted to accommodate the classification requirements. The results of this effort are highly encouraging. In the domain of CT image recognition, the top-performing model achieved a remarkable accuracy of 93.98%. In the concurrent task of categorizing cough audio data, the best performance reached 81.8%. These findings underscore the capability of this approach as an effective supplementary tool for medical diagnostics. In the face of COVID-19’s persisting impact, such machine learning advancements could significantly aid in swift and reliable diagnoses, ultimately contributing to the global battle against the pandemic.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230732

    Logistics warehousing system shelving station route planning based on X-ARM

    The need for automated control in warehouse and logistics systems has increased recently, particularly in the area of cargo storage. The use of robot arms in manufacturing is an important area for development. Multi-arm actions, including switching between multiple arms, expand the number of potential operations that numerous operators can do, but they also bring more computing hurdles. Multiple robotic arm systems must operate with careful path planning. In this study, Based on previous research papers, we propose the utilization of dRRT* and mmdRRT* algorithms for efficient path planning. The dRRT* algorithm improves exploration and convergence to an ideal path by combining the advantages of the rapidly-exploring random tree (RRT) with the optimal RRT* algorithm. To address complicated and dynamic situations, the mmdRRT* algorithm, on the other hand, uses a multi-modal distribution model. We want to improve the reliability and effectiveness of path planning for various robotic arm systems by combining these two techniques.

  • Open Access | Article 2024-02-22 Doi: 10.54254/2755-2721/41/20230735

    An empirical study on how emotion affects the probability of replies based BERT

    The rapid development of the internet, social media, and online forums have become crucial platforms for people to express their views and emotions. Comments are not only a way for users to express their opinions but also play a vital role in promoting discussions and interactions between users, significantly influencing public opinion. This paper aims to explore the impact of emotions on the likelihood of comments receiving replies, deepening the understanding of the role of emotional factors in interactions on social media and online forums. Through large-scale model training and pipeline parallel computation, this paper employs the Bidirectional Encoder Representations from Transformers (BERT) model for learning and prediction, enhancing accuracy and efficiency. The experimental results show that the response rate of negative emotional comments is about 27%, while the response rate of positive emotional comments is about 18%. It means that the comments with negative emotions are more likely to receive replies than those with positive emotions.

Copyright © 2023 EWA Publishing. Unless Otherwise Stated