Browse Items (11855 total)
Sort by:
-
Systematic Contemplate Paradigm on Diabetes Mellitus using different Machine Learning Predictive Techniques
As the foodies love fast food, from micro to combined families across the world the ratio of family members 1:4 is affected with silent killer named as diabetes. A very high blood glucose levels, metabolism, improper carbohydrate, damaged hormone insulin alleviating a human body disability leading to the silent killer of the body parts is the diabetes. An estimated 425 million of people around the globe suffering with diabetes up to 108 million to 1.7 trillion will be affected with diabetes. Therefore millennium, the universe ubiquity suffering with diabetes has next to quadrupled, growing from 9 percent and above among the people. As the eating habits of people in this trendy 21st century is dramatically devastating to the risk of overweight or obese. The silent killer diabetes consequences include kidney failure, Diabetic retinopathy, Heart attack, Stiffness of body muscles, Nerves stroke and lower limb amputation leads to type I and type II diabetes. As the researchers across the globe are using the machine learning algorithms as the reliable problem solver, The complications still continue. The purpose of this percu is to help with the apt selection of features garnishing with machine learning paradigm techniques in selecting the accurate attributes for each person to be properly diagnosed. In this archetype survey paper, we have done a systematic review chronologically a decade research which will help the researchers to explore and get the contemplate on various tangible and intangible data sets they can adopt in diagnosing the mellitus diabetes. Grenze Scientific Society, 2023. -
Pertaining analysis of fracture risk in Osteoporotic patients using Machine Learning Techniques
Bone fractures in the spine or hip are the most severe complications of Osteoporosis. Older subjects with Osteoporosis are vulnerable to falls. This paper aims to review the breakthrough in machine learning methods over the past four years in assessing fracture risk in osteoporotic patients. Machine learning is applied in the healthcare and medical field. Machine learning professionals can accurately predict disease onset by analyzing a large amount of data. Osteoporosis is one of the healthcare domains in which new Machine learning and Artificial Intelligence techniques can be implemented. The objective of this research is to give an overview of the recent advancements in machine learning methods in finding out the risk factors for fractures or predicting the onset of disease. A systematic search was conducted in PubMed to get research papers published on Machine learning methods to detect, classify, or predict osteoporosis-related fracture risk. The articles belonging to Fracture prediction and risks (n=14), Osteoporosis classification(n=3), Diagnosis of fracture(n=3), and Predicting length of stay (n=1) were identified. The quality of the articles is assessed. Most articles described the efforts to create the model and showcased excellent results in predicting the risks. Significant limitations were in the form of inadequate data splitting and data validations. More validation studies are needed in various large groups to improve the model. Most of the participants in significant studies were in their initial stage of the disease, and the reproducibility analysis was done with major disease issues. 2023 IEEE. -
Calibration of Optimal Trigonometric Probability for Asynchronous Differential Evolution
Parallel optimization and strong exploration are the main features of asynchronous differential evolution (ADE). The population is updated instantly in ADE by replacing the target vector if a better vector is found during the selection operation. This feature of ADE makes it different from differential evolution (DE). With this feature, ADE works asynchronously. In this work, ADE and trigonometric mutation are embedded together to raise the performance of an algorithm. The work finds out the best trigonometric probability value for asynchronous differential evolution. Two values of trigonometric mutation probability (PTMO) are tested to obtain the optimum setting of PTMO. The work presented in this paper is tested over a number of benchmark functions. The benchmark functions results are compared for two values of PTMO and discussed in detail. The proposed work outperforms the competitive algorithms. A nonparametric statistical analysis is also performed to validate the results. 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
Encoder-Decoder Approach toward Vehicle Detection
Vehicle Detection algorithms run on deep neural networks. But one problem arises, when the vehicle scale keeps on changing then we may get false detection or even sometimes no detection at all, especially when the object size is tiny. Then algorithms like CNN, fast-RCNN, and faster-RCNN have a high probability of missed detection. To tackle this situation YOLOv3 algorithm is being used. In the codec module, a multi-level feature pyramid is added to resolve multi-scale vehicle detection problems. The experiment was carried out with the KITTI dataset and it showed high accuracy in several environments including tiny vehicle objects. YOLOv3 was able to meet the application demand, especially in traffic surveillance Systems. Grenze Scientific Society, 2023. -
Deep Learning Algorithms Comparison forMultiple Biological Sequences Alignment
In this paper, deep learning algorithms are compared for aligning multiple biological molecular sequences such as DNA, RNA, and protein. Efficient algorithms are necessary for sequence alignment to identify significant insights, but there is a trade-off between time and accuracy. This study compares deep learning algorithms for multiple sequence alignment with better accuracy, using a new similarity measure to choose the best resemblance sequences in a set. Using a benchmark dataset, the algorithms compared include CNN, VAE, MLPNN, DBNs, Deep Boltzmann Machine, and GAN. 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
A Video Surveillance-based Enhanced Collision Prevention and Safety System
Road traffic crashes that result in fatalities have become a global phenomenon. Therefore, it is imperative to use caution and vigilance while being on the road. Human mistake, going over the speed limit, being preoccupied while driving or walking, disobeying safety precautions, and other factors can also contribute to such unforeseen accidents or injuries, which can result in both bodily and material loss. So, safety is what we seek to achieve. Furthermore, as the number of automobiles has increased, so too have collisions between vehicles and pedestrians. Using computer vision and deep learning approaches, this research seeks to anticipate such encounters. The data often comes from traffic surveillance cameras in video formats. We have therefore concentrated on video sequences of vehicle-pedestrian collisions. We begin with a detection phase that includes the identification of vehicles and pedestrians; for this phase, we employed YOLO v3 (You Only Look Once). YOLO v3 has 80 classes, but we only took six of them: person, car, bike, motorcycle, bus, and truck. Following detection, the Euclidean distance approach is used to determine the interspace between the vehicle and the pedestrian. The closer the distance between a vehicle and a pedestrian, the more likely it is that they will collide. As a result, pedestrians in risk are located, and once we are aware of the pedestrians in danger, we search for nearby safer regions to alert them to head to the nearest location that is secure. Grenze Scientific Society, 2023. -
Analytical Results of Heart Attack Prediction Using Data Mining Techniques
In the modern era of living a fast lifestyle, people are not more conscious of their food eating and lifestyle. Due to these reasons, the chances of having a cardiac-related disease have risen drastically. This paper has studied the various supervised and unsupervised machine learning algorithms in comparative methods with best accuracy. Models like classification algorithms, regression algorithms, and clustering algorithms have been used for this paper. This research paper majorly focuses on patients with certain medical attributes that indicate a higher risk of heart disease. The model almost gives a good accuracy for all the regression and classification models when compared to the clustering models. Among all the algorithms, random forest and decision tree gives better accuracy 2023 IEEE. -
A Space Vector Modulated Direct Torque Control of Induction Motor with Improved Transient Performance and Reduced Parameters Dependency
Direct torque control (DTC) of induction motors is hampered by high torque and current ripple. Integrating DTC with space vector pulse width modulation (DTC-SVPWM) is one of the frequently used approaches to solve this problem. However, it adds to the computational complexity, increases the number of necessary motor parameters needed for control scheme implementation, and also affects the transient performance of the induction motor; this approach compromises the robustness and simplicity of DTC scheme. To get around these restrictions, a novel control strategy is put forth in this paper. The suggested scheme enhances the steady-state performance and transient response of the motor while preserving the simplicity and robustness of the DTC scheme. To accomplish this, the proposed control scheme operates at varying switching frequencies during transient conditions and constant switching frequencies during steady-state. The suggested speed control method does not employ any rotating reference frame transformations or usage of many rotor parameters for computation, nor does it call for sector identification and operates with a single PI controller. The suggested topology also uses a bus-clamped PWM modulation technique, which lowers the average switching frequency to 2/3 times the actual switching frequency. Thus, switching losses are also decreased. Simulation results show the effectiveness of the proposed topology in enhancing the transient and steady-state performance of the induction motor. The results are compared with the traditional DTC and DTC-SVPWM scheme. 2023 IEEE. -
Kho Kho Model: A Novel Technique for Efficient Handoff in Vehicular Ad-hoc Networks
The highly mobile nature of VANET implies that the nodes involved are constantly disconnecting and reconnecting as they switch between access points or move out of the range of their access points. In such scenarios, seamless connectivity is essential, especially when emergency services are involved. Handoff is a process in wireless communication that takes care of the switching process that happens between access points whenever a mobile device moves from one point to another. In a dynamic scenario involving vehicular nodes, this switching needs to take place between a mobile node or a fixed access point (known as RSUs), as quickly as possible. To this end, this research work proposes a novel handoff method known as the Kho Kho Model - which is loosely based on the traditional Indian sport of the same name. The model groups together nodes that are moving in the same direction, thereby effectively reducing the amount of processing required to perform handoff for a set of nodes. The use of ANN have helped to improve handoff since it can help in making decisions quickly by making use of multiple parameters including signal strength, noise, direction, and others. To improve the efficiency of the proposed handoff model, RBFNN has been used in this research. The proposed model was implemented using NS-3 simulator. The results have shown that the proposed method has a slightly better improvement in the overall NRO, a reduced average delay and reduced jitter compared to the existing handoff method employed by the IEEE 802.11p standard. 2023 IEEE. -
Performance Evaluation of Convolutional Neural Networks for Stellar Image Classification: A Comparative Study
This study analyzes three distinct convolutional neural network (CNN) models, ResNet, Parallel CNN, and VGG16, for object classification using the Star-Galaxy Classification dataset. The dataset comprises a vast collection of celestial object images, including galaxies, stars, and quasars. The effectiveness of each CNN model is evaluated based on accuracy, a commonly used performance metric. The results reveal that the Parallel CNN model achieved the highest accuracy of 90.08% in classifying celestial objects, followed by VGG16 with an accuracy of 86%, and ResNet with an accuracy of 83%. Specifically, the Parallel CNN model demonstrates superior performance in classifying galaxies and stars. These findings provide valuable insights into the strengths and weaknesses of each model for this specific classification task, guiding the development of more effective CNN models for similar applications in cosmology and other fields. This research contributes to the growing literature on CNN models' application in astronomy and underscores the importance of selecting appropriate models to achieve high accuracy in object classification tasks. The study's insights can be utilized to inform the development of more effective CNN models for similar tasks and facilitate advancements in astronomical research. 2023 IEEE. -
Artificial Intelligence Involvement in Graphic Game Development
Games have always been a popular form of entertainment and with the advancements in technology, the integration of Artificial Intelligence (AI) in gaming has revolutionized the gaming industry. This research article aims to explore the various applications of AI in gaming and its impact on the industry and player experience. Unlike the typical straightforward nature of AI, this research paper takes a more human approach to discussing the topic. It delves into the evolution of AI in games and the various types of AI used in game development. These include rule-based AI, learning- based AI, and evolutionary AI, which have all contributed to the development of increasingly immersive gaming experiences. The benefits and challenges of using AI in games are also explored, considering the impact on player experience. While AI-powered opponents can provide a greater challenge, balancing the difficulty level is critical to ensuring the game remains enjoyable. The potential ethical concerns of using AI in games are also discussed, such as data privacy, bias, and fairness. Furthermore, this research paper looks into the future of AI in games and how it may shape the gaming industry and player experience in the years to come. With the continued development of AI techniques such as reinforcement learning and GANs, the possibilities for more immersive and engaging gaming experiences are endless. 2023 IEEE. -
HumanComputer Interactions with ArtificialIntelligence and Future Trends of HCIA Study
Artificial Intelligence, the name itself depicts the meaning that providing the knowledge of human to the machine artificially. AI is not a sense or feeling but the software or a model evolved to do complex tasks like human beings. With the invention of computer it has become so easy to do day to day jobs without much effort. HCI is all about interacting with computers. Now-a days it is possible to mesh with the computer through voice, touch, eye movement, and hand gestures. HCI has many challenges but has established in grand manner with the support of Artificial Intelligence. This study provides some important roles of Artificial Intelligence in HCI and its future development. 2023, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. -
Data Analysis on Hypothyroid Profiles using Machine Learning Algorithms
Machine learning algorithms enable computers to learn from data and continuously enhance performance without explicit programming. Machine learning algorithms have significantly improved the accuracy and efficacy of thyroid diagnosis. This study identified and analysed the usefulness of several machine-learning algorithms in predicting hypothyroid profiles. The main goal of this study was to see the extent to which the algorithms adequately assessed whether a patient had hypothyroidism. Age, sex, health, pregnancy, and other factors are among the many factors considered. Extreme Gradient Boosting Classifier, Logistic Regression, Random Forest, Long-Term Memory, and K-Nearest Neighbors are some of the machine learning methods used. For this work, two datasets were used and analysed. Data on hypothyroidism was gathered via DataHub and Kaggle. These algorithms were applied to the collected data based on metrics such as Precision, Accuracy, F1 score and Recall. The findings showed that the Extreme Gradient Boosting classification method outperformed the others regarding F1 score, accuracy, precision, and recall. The research demonstrated how machine learning algorithms might predict thyroid profiles and identify thyroid-related illnesses. 2023 IEEE. -
Diabetic Retinopathy Diagnosis Using Retinal Fundus Images through MobileNetV3
Diabetic Retinopathy (DR) is a major disease throughoutthe world. Diagnosis of diabetes at an early stage is so critical and could help save several lifestyles. One out of two individuals experiencing diabetes has been determined to have some phase of DR. Recognition of DR symptoms in time can turn away the vision weakness inmost the cases, nonetheless, such disclosure is troublesome with present devices and strategies. Existingmethods for determining whether a person is suffering from diabetes or maybe the chances of acquiring diabetesrely heavily on examiners. Most of the time, it can be treated if caught during the early stages. There is a need for creating models that are efficient and robust to detect DR holistically. In recent times the advent of Deep learning models has been used extensively in various Bio medical applications. In this work, we utilize a Hyper parameter tuned MobileNet-V3 model based on a multi-stage Convolutional Neural Network (CNN) to efficiently classify images from the IRDID dataset. A Multiclass classification model involving images collated from various sources were trained, validated and tested for classification accuracy. The network was evaluated based on parameters and the network was able to achieve an accuracy of 88.6% 2023 IEEE. -
Data Analysis and Machine Learning Observation on Production Losses in the Food Processing Industry
Food wastage and capturing lineage from production to consumption is a bigger concern. Yielding, storage and transportation areas have evolved to a great extent associated to manufacturing and automation which lead to technical advancements in food processing industry. In such situation, losses are generally observed in the crop production which are sometimes minimal and ignored. However, in some cases these losses are huge and are becoming a threat to the both producers and consumers. Here we considered data related to dairy products and analysed the production losses especially while processing them in the treating unit. Literature on parameters and associated data analysis in the form of graphical representation are provided in the appropriate sections of the paper. Linear regression and correlation were envisaged in view of incorporating machine learning techniques understanding production losses. Karl Pearson's correlation provides an observation related to association of parameters which are desired to be less coupled in terms of employing proposed newer methodology. 2023 IEEE. -
Brain Tumor Classification: A Comparison Study CNN, VGG 16 and ResNet50 Model
Brain tumors pose a severe threat to global health and may be lethal. Early detection and classification of brain tumors are essential for successful therapy and better patient outcomes. The good news is that advances in deep learning techniques have shown tremendous promise in medical image analysis, particularly in the detection and classification of brain tumors. Convolutional Neural Networks (CNN), a class of deep learning models, are used to process and analyze visual input, notably images, and movies. They excel in computer vision tasks like object detection, image segmentation, and categorization. Popular and efficient image analysis methods include CNNs. VGG 16 and ResNet 50 are two examples of deep convolutional neural network architectures used for image categorization applications. A number of image identification problems have been successfully solved using the 16 layer VGG 16. ResNet50, a well known 50 layer architecture, employs residual connections to get over the vanishing gradient issue and permits the training of deeper networks. A proprietary CNN model, VGG 16, and ResNet50 were compared in studies to see how well they performed on a dataset. The VGG 16, ResNet50, and the tailored CNN model were the most precise models. As a consequence, VGG 16 accurately detects brain cancers in the dataset that was supplied. Overall, this study highlights the value of deep learning techniques for medical image processing and their potential to improve the accuracy and efficacy of brain tumor diagnosis and treatment. 2023 IEEE. -
Marketing Research and Market-Focused Production as an Effective Business Tool in Power Sector
Businesses must devote part of their resources to conducting market and marketing research to make good decisions, which will help expand any business and utilize resources effectively. Understanding the intended clients is essential to successfully operating and expanding a firm. For marketers to comprehend consumer value about the product being supplied and therefore add value to their consumers, it is crucial to have this understanding. Organizations can better influence customers to buy niche goods or corporate services after thoroughly understanding their objectives, requirements, and values. In this situation, it is required to restructure the physical system and the related control and planning systems to provide production the tools it needs to become more competitive and customer-focused, acting as a positive and active production process instead of a reactive one. One of the finest techniques for understanding consumers is market research. It provides basic information that a company may utilize to inform its marketing strategy, facilitating and enhancing sales and marketing. This paper reviews the impact of effective market and marketing research and market-focused manufacturing in the power sector. 2023 IEEE. -
Facial Emotion Detection Using Deep Learning: A Survey
The long history of facial expression analysis has influenced current research and public interest. The scientific study and comprehension of emotion are credited to Charles Darwin's 19th-century publication The Representation of the Sentiment in Man and Animals (originally published in 1872). As Recognition of human emotions from images is one of the utmost important and difficult societal connection study assignments. One advantage of using a deep learning strategy is its independence from human intervention while undertaking feature engineering. This approach involves an algorithm that scans the data for features that connect, then combines them to promote quicker learning without being explicitly told to. Deep learning (DL) based emotion detection outperforms traditional image processing methods in terms of performance. In this analytical study, the creation of an artificial intelligence (AI) system that can recognize emotions from facial expressions is presented. It discusses the various techniques for doing so, which generally involve three steps: face uncovering, feature extraction, and sentiment categorization. This study describes the various existing solutions and methodologies used by the researchers to build facial landmark interpretation. The Significance of this survey paper is to analyze the recent works on facial expression detection and distribute better insights to novice researchers for the upgradation in this domain. 2023 IEEE. -
Enhanced Design and Performance Analysis of a Seven-Level Multilevel Inverter for High-Power Applications
The structure and performance analysis of a seven-level multilevel inverter is discussed in this study. Due to their capacity to get around the drawbacks of traditional two-level inverters, like high voltage stress on power devices and harmonic distortion, multilevel inverters have attracted a lot of attention lately. Multiple voltage levels can be produced by the seven-level multilevel inverter which is being proposed because it uses a sequential arrangement of power sources and capacitors. The design methodology involves selecting appropriate power devices and capacitance values to achieve the desired voltage levels while minimizing losses and ensuring reliable operation. Total harmonic distortion (THD), inverter efficiency, and voltage stress on power devices are all considered as part of the performance analysis. In comparison to conventional two-level inverters, simulation results indicate that the proposed seven-level multilevel inverter offers lower THD, increased efficiency, and reduced voltage stress. This research contributes to the advancement of multilevel inverter technology and its potential applications in various power conversion systems. 2023 IEEE. -
Student's Performance Prediction Using Modified PSO
In the present-day education system, evaluating students' performance is very important. The prediction of student performance is helpful to students and instructors in maintaining a student's progress. Presently, institutions are implementing Continuous Evaluation Systems. This system is beneficial for improving the students' performance and helps increase their regularity in their studies. This work proposed a new version of the particle swarm optimization algorithm to classify students into different categories based on their performance with the help of various factors. The proposed model modified the strategy to increase the optimization algorithm's efficiency. This model deals with the bag of features with the modified optimization algorithm to recognize students' performance based on 33 features. These features are used for training, validation, and testing of the student's performance and generate a classification report which includes F-measure, recall, precision, and accuracy. Based on the classification report, the proposed model is compared with the existing methods, and the better performance of the proposed model is ascertained. 2023 IEEE.