Yazar "Koyuncu, Hakan" seçeneğine göre listele
Listeleniyor 1 - 20 / 24
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe A comparative study of handwritten character recognition by using Image Processing and Neural Network techniques(2021) Koyuncu, HakanThis study aims to analyze the effects of noise, image filtering, and edge detection techniques in the preprocessing phase of character recognition by using a large set of character images exported from the MNIST database trained with various sizes of neural networks. Canny and Sobel algorithms are deployed to detect the edges of the images. The Canny algorithm can produce smoother and thinner continuous edges compare to the Sobel algorithm. The structural forms were reshaped using the Skeletonization algorithm. The Laplacian filter was used to increase the sharpness of the images and High pass filtering was used to highlight the fine details in blurred images in the form of image filtering. Gaussian noise or image noise with Gaussian intensity was used in Matlab on MNIST character images with the probability density function P. The effects of noise on character images are displayed during character recognition related to Neural network properties. Neural networks are commonly used to recognize patterns among optical characters. Feedforward neural networks are deployed in this study. A comprehensive analysis of the image processing algorithms is included during character recognition. Improved accuracy is observed with character recognition during the prediction phase of the neural networks. A sample of unknown numeric characters is tested with the application of High pass filtering plus feedforward neural network and 89% average output prediction accuracy was obtained against the average number of hidden layers in the neural network. Other prediction accuracies were also tabulated for the reader’s attention.Öğe A Comprehensive Survey of Predicting stock market prices: an analysis of traditional statistical models and machine-learning techniques(American Institute of Physics, 2024) Patel, Divy; Patel, Warish; Koyuncu, HakanThe stock market has witnessed a remarkable surge in popularity in recent years, attracting investors from all walks of life. However, predicting stock values remains a daunting task due to financial markets' inherent unpredictability and complexity. Despite these challenges, the stock market offers a dynamic and ever-changing platform for traders to invest in shares, with the potential for significant gains and losses. For investors, accurate forecasting of stock prices is crucial as it provides invaluable insights into a company's financial health and growth prospects. With this information, investors can make informed decisions, mitigate risks, and capitalize on lucrative opportunities in the market. As a result, extensive research has been dedicated to developing effective prediction methods, leveraging various mathematical models and machine-learning techniques. This research paper delves into the realm of stock market prediction, explicitly focusing on evaluating different machine-learning styles. The primary objective is to comprehensively analyze and compare the performance of these techniques in forecasting stock market behavior. By understanding the strengths and limitations of each method, investors, financial analysts, and market participants can gain critical knowledge to optimize their trading strategies and decision-making processes. To achieve this goal, the study explores an array of machine-learning algorithms, ranging from traditional linear regression models to sophisticated deep-learning approaches. These algorithms leverage historical stock market data, macroeconomic indicators, company financials, and sentiment analysis, among other factors, to predict future price movements and market trends. In addition to performance comparison, the research paper examines the impact of various factors that influence the effectiveness of these machine-learning techniques. Factors such as data quality, feature engineering, model selection, hyperparameter tuning, and market conditions play pivotal roles in the accuracy of predictions. Understanding these factors will aid in refining the model-building process and enhancing overall forecasting capabilities. The study encompasses an extensive dataset spanning multiple stock markets and periods, ensuring robustness and reliability in the findings. Performance evaluation metrics, including mean squared error, accuracy, precision, recall, and F1 score, will be employed to assess the predictive power of the machine-learning techniques objectively. Furthermore, the paper investigates the potential of ensemble methods, combining the strengths of multiple models to achieve enhanced prediction accuracy. Ensemble techniques, such as bagging, boosting, and stacking, have proven effective in diverse domains and are expected to demonstrate their value in stock market prediction. By the end of this research, readers will have a comprehensive understanding of the landscape of machine-learning techniques applied to stock market prediction. The findings will offer insights into which methods are most suitable for different market conditions and will aid in establishing best practices for effective and reliable stock market forecasting. In conclusion, this research paper serves as a valuable resource for investors, financial analysts, and researchers, thoroughly assessing machine-learning techniques' efficacy in predicting stock market behavior. It contributes to the growing body of knowledge in financial technology. It underscores the critical role of data-driven decision-making in navigating the complexities of the modern stock market.Öğe A deep learning approach to differentiate between acute asthma and bronchitis in preschool children(Institute of Electrical and Electronics Engineers Inc., 2022) Salih, Waleed; Koyuncu, HakanAcute asthma and bronchitis are common diseases and spread rapid that affecting children all around the world, especially in preschool children (under six years), most people confuse between the two diseases due overlapping symptoms, where it is difficult to diagnose cases by junior doctors in the hospitals and the wrong diagnosis sometimes leads to death, the main objective of this study is the right diagnosis of cases and differentiation between them, to preserve the lives of children and reduce the spread of disease which saves effort, time and money for the health institutions, in this study we will present two deep learning models for binary classification of real dataset that collected in the teaching hospital for children afflicted, which was examined by a pediatrician consultant, the final results showed that the convolutional neural network model (CNN) outperformed with 99.350% accuracy to the second classifier the long short-term memory model (LSTM), thus the CNN was adopted as a binary classification model for our study.Öğe A hybrid ensemble learning approach for efficient diabetic retinopathy prediction and classification using machine learning and deep learning techniques(Ismail Saritas, 2024) Shah, Arpit; Patel, Warish; Koyuncu, HakanAI is a crucial tool in early detection and classification of diabetic retinopathy, which is a leading cause of visual impairment globally. Transfer Learning (TL) was used to improve the accuracy of predictions and classifications within training datasets, surpassing existing methodologies. The study provides comprehensive insights into current databases, screening programs, performance evaluation metrics, relevant biomarkers, and challenges encountered in ophthalmology. The findings underscore the potential of AI-based approaches in enhancing diagnostic precision and offer a promising direction for future studies. The paper concludes by delineating opportunities for further research and development in integrating AI advancements in the field. Conclusion: The findings underscore the efficacy of Transfer Learning in significantly improving the accuracy of diabetic retinopathy image predictions. This research highlights the potential of AI-based approaches in enhancing diagnostic precision and offers a promising direction for future studies. The paper concludes by delineating opportunities for further research and development, emphasizing the continued integration of advanced AI methodologies in ophthalmology to advance diabetic retinopathy detection and management.Öğe A novel deep learning framework enhanced by hybrid optimization using dung beetle and fick’s law for superior pneumonia detection(Multidisciplinary Digital Publishing Institute (MDPI), 2024) Sabaawi, Abdulazeez M.; Koyuncu, HakanPneumonia is an inflammation of lung tissue caused by various infectious microorganisms and noninfectious factors. It affects people of all ages, but vulnerable age groups are more susceptible. Imaging techniques, such as chest X-rays (CXRs), are crucial in early detection and prompt action. CXRs for this condition are characterized by radiopaque appearances or sometimes a consolidation in the affected part of the lung caused by inflammatory secretions that replace the air in the infected alveoli. Accurate early detection of pneumonia is essential to avoid its potentially fatal consequences, particularly in children and the elderly. This paper proposes an enhanced framework based on convolutional neural network (CNN) architecture, specifically utilizing a transfer-learning-based architecture (MobileNet V1), which has outperformed recent models. The proposed framework is improved using a hybrid method combining the operation of two optimization algorithms: the dung beetle optimizer (DBO), which enhances exploration by mimicking dung beetles’ navigational strategies, and Fick’s law algorithm (FLA), which improves exploitation by guiding solutions toward optimal areas. This hybrid optimization effectively balances exploration and exploitation, significantly enhancing model performance. The model was trained on 7750 chest X-ray images. The framework can distinguish between healthy and pneumonia, achieving an accuracy of 98.19 ± 0.94% and a sensitivity of 98 ± 0.99%. The results are promising, indicating that this new framework could be used for the early detection of pneumonia with a low cost and high accuracy, especially in remote areas that lack expertise in radiology, thus reducing the mortality rate caused by pneumonia.Öğe Advancements in telehealth: enhancing breast cancer detection and health automation through smart integration of IoT and CNN deep learning in residential and healthcare settings(Semarak Ilmu Publishing, 2025) Duodu, Nana Yaw; Patel, Warish D.; Koyuncu, HakanThe rapid evolution of telehealth, or telemedicine, has spurred crucial technological advancements aimed at addressing the early stages of complex cancer conditions, where conventional diagnostic methods face challenges. This research introduces a cancer detection system that utilizes Internet of Things (IoT)-based patient records and machine learning. The primary objective is to automate real-time breast cancer monitoring and detection in residential institutions and smart hospitals, thus enhancing the delivery of quality cancer healthcare. Background: Traditional diagnostic methods, particularly physical inspection, exhibit inherent limitations in identifying breast cancer at early stages. This research responds to this challenge by leveraging innovative technologies, such as IoT and deep learning-based techniques, to overcome the constraints of conventional approaches. Objective: The primary goal of this study is to develop and implement a cancer detection system that integrates IoT-based patient records and machine learning for real-time breast cancer monitoring in residential and healthcare settings. Method: The research employs a synergistic combination of IoT technology for collecting images of residential users and Convolutional Neural Network (CNN), a deep learning technique, for early cancer prediction. The focus lies on contributing to the overall well-being of individuals who may unknowingly be living with cancer. Result: Simulated outcomes after 25 epochs are presented, emphasizing the training accuracy of the model and its validation accuracy using the proposed VGG16 classifier. Graphical representations of the results indicate consistent performance metrics, with both validation and training accuracy exceeding 99%. Specifically, the training accuracy measures at an impressive 99.64%, while the validation accuracy stands at 99.12%. Main Findings: The study demonstrates the effectiveness of the integrated IoT and deep learning techniques in achieving high accuracy rates for early breast cancer prediction. The findings affirm the potential of this approach to assist dermatologists in identifying breast malignancies at treatable stages. Conclusion: This research establishes a foundational framework for the integration of IoT and deep learning techniques, presenting a promising avenue for advancing early cancer detection in smart healthcare systems. The proposed cancer detection system holds significant potential for improving healthcare outcomes and contributing to the overall well-being of individuals at risk of breast cancer.Öğe Construction of 3D soil moisture maps in agricultural fields by using wireless sensor communication(Gazi Univ, 2021) Koyuncu, Hakan; Gündüz, Burak; Koyuncu, BakiOver-irrigation without considering the soil property reduce the product yield and variety in many agricultural areas. In this study, it is aimed to produce a more useful, and user-friendly 3D soil moisture detection system by using wireless communication across the agricultural areas. The deficiencies of agricultural land can be eliminated in terms of irrigation, product variety, and product yield. 3D moisture information obtained from the soil can be transferred to a database system and the farmers can use this system to cultivate across the correct fields. A capacitive soil moisture sensor is deployed as a sensor unit. Each sensor unit with its electronics is placed in a PVC pipe with a specific length. This PVC pipe is placed vertically in the soil with sensor electrodes contacting the soil. Moisture measurements are carried out across the agricultural area. The system provides 3D moisture maps of the soil at fixed depths. Each 3D map represents a subsurface moisture layer. The sensor units are calibrated by measuring the moisture in the water, corresponding to %100 moisture in the soil, and the moisture in dry air, corresponding to %0 moisture in the soil. A percentage moisture determination formula is developed between these two extreme levels for each sensor unit. Hence the benefit of the results will be the knowledge of % moisture values in-depth profile of the agricultural areas. Farmers will have comprehensive and real-time information about moisture data and this data will help them to grow better crops.Öğe Deep learning and grey wolf optimization technique for plant disease detection: a novel methodology for improved agricultural health(International Information and Engineering Technology Association, 2023) Jabbar, Amenah Nazar; Koyuncu, HakanPlant disease outbreaks have a profound impact on the agricultural sector, leading to substantial economic implications, compromised crop yields and quality, and potential food scarcity. Consequently, the development of effective disease prevention and management strategies is crucial. This study introduces a novel methodology employing deep learning for the identification and diagnosis of plant diseases, with a focus on mitigating the associated detrimental effects. In this investigation, Convolutional Neural Networks (CNNs) were utilized to devise a disease identification method applicable to three types of plant leaves - peppers (two classes), potato (three classes), and tomato (nine classes). Preprocessing techniques, including image resizing and data augmentation, were adopted to facilitate the analysis. Additionally, three distinct feature extraction methods - Haralick feature, Histogram of Gradient (HOG), and Local Binary Patterns (LBP) - were implemented. The Grey Wolf Optimization (GWO) technique was employed as a feature selection strategy to identify the most advantageous features. This approach diverges from traditional methodologies that solely rely on CNNs for feature extraction, instead extracting features from the dataset through multiple extractors and passing them to the GWO for selection, followed by CNN classification. The proposed method demonstrated high efficiency, with classification accuracies reaching up to 99.8% for pepper, 99.9% for potato, and 95.7% for tomato. This study thus provides a progressive shift in plant disease detection, offering promising potential for improving agricultural health management. In conclusion, the integration of deep learning and the Grey Wolf Optimization technique presents a compelling approach for plant disease detection, demonstrating high accuracy and efficiency. This research contributes a significant advancement in the field of agricultural health and disease management.Öğe Effective audio and video-based learning for youngest age(2024) Akram, Basmal Omer; Koyuncu, HakanThe research project is an interactive educational book and it is a game, but in reality it is a measure of intelligence through which intelligence breakthroughs are achieved through the accuracy of answering questions with the speed of answering. The focus was on children under the age of six, but it included a number of those who are older. Discussing through research on the types of education, focusing on interactive and electronic education, explaining the difference between the two, explaining the types of human intelligence, how to build a school system through the use of a calculator, and what are the requirements and types for that, leading to the importance of the impact of the diversity of educational media, the importance of learning through play, and the benefits and harms of each case. Reaching results indicating that students prefer studying in class to studying electronically, it was noted that there are mental mutations for every age, but the age that responded most and most quickly was the age of youth and advanced youth. The reasons for this result were explained through the research and what we reached, and some methods were presented to increase mental efficiency.Öğe Empowering Health and Well-being: IoT-Driven Vital Signs Monitoring in Educational Institutions and Elderly Homes Using Machine Learning(Forex Publication, 2024) Duodu, Nana Yaw; Patel, Warish D.; Koyuncu, Hakan; Nartey, Felix; Torgby, WisdomIoT-based EHRs use machine learning technology to automate real-time patient-centered records more securely for authorized users. Background: In this era of pandemics, predictive healthcare systems are necessary for private and public healthcare delivery to predict early cancer, COVID-19, hypertension, and fever in Educational Institutions and Elderly Homes. IoT-Based EHRs bring healthcare delivery to the doorsteps of educational home facilities users, thereby reducing the time required to access healthcare and minimizing direct physical interaction between individuals seeking healthcare and their providers. Method: This research work proposed a real-time intelligent IoT-based EHR system that generates vital signs of students within the educational environment using contactless sensors (Raspberry Pi Noir Camera, rPPG camera) and contacted wearable sensors composed of enzymatic sensor, immunogens, and Nano sensors to detect cancer (Leukaemia). AFTER CAPTURING THE PHYSIOLOGICAL DATA, THE in-build EWS plots system determines the condition and further triggers the criticality (abnormality) in health status. Discussion: For effective health status prediction by the proposed plan, the vital sign dataset was used to train a model for the proposed method. Among the best-performing models, the random forest algorithm proved a better model, with an accuracy of 99.66% and an error rate of 0.34%. Conclusion: The Home HMS seeks to improve health prediction in institutional homes for users' overall well-being. © 2024 by the Nana Yaw Duodu, Warish D. Patel, Hakan Koyuncu, Felix Nartey, Wisdom Torgby.Öğe Empowering healthcare innovation: IoT-enabled smart systems and deep learning for enhanced diabetic retinopathy in the telehealth landscape(Taru Publications, 2024) Shah, Arpit; Patel, Warish; Koyuncu, HakanBackground: Among prevalent medical complications, Diabetic Eye Disease (DED) stands as a significant contributor to vision loss. To forecast its progression and accurately assess the various stages, diverse methodologies have emerged. Machine Learning (ML) and Deep Learning (DL) algorithms have become essential tools in this endeavor, primarily through their adept analysis of Diabetic Retinopathy (DR) images. However, there is still a need for a more efficient and accurate method to predict DR performance. Method: We have developed an innovative method for classifying and predicting diabetic retinopathy. The novel idea in this research is to combine several techniques, including ensemble learning and a 2D convolutional neural network; we utilized transfer learning and a correlation method in our approach. Initially, the Stochastic Gradient Boosting process was employed for predicting diabetic retinopathy. We then used a boosting-based Ensemble Learning method for predicting images of diabetic retinopathy. Next, we applied a 2D Convolutional Neural Network. We successfully employed Transfer Learning to classify different stages of diabetic retinopathy images accurately. This research explores the role of artificial intelligence in identifying and categorizing diabetic retinopathy at an early stage, using techniques such as machine learning and deep learning. It also use techniques like transfer learning, domain adaptation, multitask learning, and explainable AI to accurately classify different stages of diabetic retinopathy images. Our proposed technique achieves impressive results through experiments, with a 97.9% accuracy in forecasting DR images and a 98.1% accuracy in image grading. Additionally, sensitivity and specificity metrics measure 99.3% and 97.6%, respectively. Comparative analysis with existing methods underscores the high predictive accuracy achieved by our proposed approach.Öğe Enhancing SDN anomaly detection: a hybrid deep learning model with SCA-TSO optimization(Science and Information Organization, 2024) Alhilo, Ahmed Mohanad Jaber; Koyuncu, HakanThe paper explores the evolving landscape of network security, in Software Defined Networking (SDN) highlighting the challenges faced by security measures as networks transition to software-based control. SDN revolutionizes Internet technology by simplifying network management and boosting capabilities through the OpenFlow protocol. It also brings forth security vulnerabilities. To address this we present a hybrid Intrusion Detection System (IDS) tailored for SDN environments leveraging a state of the art dataset optimized for SDN security analysis along with machine learning and deep learning approaches. This comprehensive research incorporates data preprocessing, feature engineering and advanced model development techniques to combat the intricacies of cyber threats in SDN settings. Our approach merges feature from the sine cosine algorithm (SCA) and tuna swarm optimization (TSO) to optimize the fusion of Long Short Term Memory Networks (LSTM) and Convolutional Neural Networks (CNN). By capturing both spatial aspects of network traffic dynamics our model excels at detecting and categorizing cyber threats, including zero-day attacks. Thorough evaluation includes analysis using confusion matrices ROC curves and classification reports to assess the model’s ability to differentiate between attack types and normal network behavior. Our research indicates that improving network security using software defined methods can be achieved by implementing learning and machine learning strategies paving the way, for more reliable and effective network administration solutions.Öğe Hand gesture recognition for interactive media player using CNN and image classification(Institute of Electrical and Electronics Engineers Inc., 2022) Awad, Anwar Diyaa; Koyuncu, HakanIn this paper the problem of gesture recognition is addressed with a focus on the recognition of handshapes. Due to the number of parameters to be considered (joint angles, hand position, three-dimensional orientation, as well as muscle and skin deformations), even the isolated problem of handshapes recognition becomes very complicated for a solution using conventional deterministic algorithms. Machine learning methods. In this paper, we evaluated the 26 signs of the Sign Language but the results can then be extended to any gesture or movement performed with only one hand. Certainly, even if these gestures can be recognized precisely.Öğe Intelligence feeder system for stray cats(Institute of Advanced Engineering and Science, 2023) Saadoon, Alaa Mohsin; Koyuncu, HakanRecently, the care of stray cats has become an important matter, given the difficulty for these animals to obtain food or water, whether in wild or remote areas, and this may cause the death of these animals. The aim of this work is to design and implement a low-cost feeding system for feral and indoor cats. The system is controlled by a jetson nano and arduino mega 2,560. The cat detection feeder system is built using the single shot multibox detector (SSD) MobileNet V2 algorithm. The system provides food and water for the cats. The SSD on jetson nano is implemented in real time. Jetson nano takes a picture using a webcam then runs the SSD algorithm. When a cat is detected, the arduino turns on a servo motor and a water pump. The system also includes a sim800l and GPS NEO-6M modules to send an alert message when the food and water tanks are empty. The message also contains the location of the feeder. When testing the system to determine the effectiveness of the functions, the SSD algorithm succeeded in recognizing cats, the system successfully provided food for the cats, and all parts of the device worked in high harmonic. © 2023 Institute of Advanced Engineering and Science. All rights reserved.Öğe Merging two models of one-dimensional convolutional neural networks to improve the differential diagnosis between acute asthma and bronchitis in preschool children(2024) Waleed, Salih; Koyuncu, HakanBackground: Acute asthma and bronchitis are common infectious diseases in children that affect lower respiratory tract infections (LRTIs), especially in preschool children (below six years). These diseases can be caused by viral or bacterial infections and are considered one of the main reasons for the increase in the number of deaths among children due to the rapid spread of infection, especially in low- and middle-income countries (LMICs). People sometimes confuse acute bronchitis and asthma because there are many overlapping symptoms, such as coughing, runny nose, chills, wheezing, and shortness of breath; therefore, many junior doctors face difficulty differentiating between cases of children in the emergency departments. This study aims to find a solution to improve the differential diagnosis between acute asthma and bronchitis, reducing time, effort, and money. The dataset was generated with 512 prospective cases in Iraq by a consultant pediatrician at Fallujah Teaching Hospital for Women and Children; each case contains 12 clinical features. The data collection period for this study lasted four months, from March 2022 to June 2022. (2) Methods: A novel method is proposed for merging two one-dimensional convolutional neural networks (2-1D-CNNs) and comparing the results with merging one-dimensional neural networks with long short-term memory (1D-CNNs + LSTM). (3) Results: The merged results (2-1D-CNNs) show an accuracy of 99.72% with AUC 1.0, then we merged 1D-CNNs with LSTM models to obtain the accuracy of 99.44% with AUC 99.96%. (4) Conclusions: The merging of 2-1D-CNNs is better because the hyperparameters of both models will be combined; therefore, high accuracy results will be obtained. The 1D-CNNs is the best artificial neural network technique for textual data, especially in healthcare; this study will help enhance junior and practitioner doctors' capabilities by the rapid detection and differentiation between acute bronchitis and asthma without referring to the consultant pediatrician in the hospitals.Öğe Novel hybrid optimization techniques for enhanced generalization and faster convergence in deep learning models: the nestyogi approach to facial biometrics(Multidisciplinary Digital Publishing Institute (MDPI), 2024) Altaher, Raoof; Koyuncu, HakanIn the rapidly evolving field of biometric authentication, deep learning has become a cornerstone technology for face detection and recognition tasks. However, traditional optimizers often struggle with challenges such as overfitting, slow convergence, and limited generalization across diverse datasets. To address these issues, this paper introduces NestYogi, a novel hybrid optimization algorithm that integrates the adaptive learning capabilities of the Yogi optimizer, anticipatory updates of Nesterov momentum, and the generalization power of stochastic weight averaging (SWA). This combination significantly improves both the convergence rate and overall accuracy of deep neural networks, even when trained from scratch. Extensive data augmentation techniques, including noise and blur, were employed to ensure the robustness of the model across diverse conditions. NestYogi was rigorously evaluated on two benchmark datasets Labeled Faces in the Wild (LFW) and YouTube Faces (YTF), demonstrating superior performance with a detection accuracy reaching 98% and a recognition accuracy up to 98.6%, outperforming existing optimization strategies. These results emphasize NestYogi’s potential to overcome critical challenges in face detection and recognition, offering a robust solution for achieving state-of-the-art performance in real-world applications.Öğe Obstacle avoidance in unmanned aerial vehicles using image segmentation and deep learning(Institute of Electrical and Electronics Engineers Inc., 2022) Obaid, Amro Ali; Koyuncu, HakanMachine learning is a branch of artificial intelligence based on the idea that systems can learn to identify patterns and make decisions with a minimum of human intervention. In this study, demonstration learning will be used, using neural networks in a prototype of a drone built to perform trajectories in controlled environments. To accelerate the training convergence process, a new training data selection approach has been introduced, which picks data from the experience pool based on priority instead of randomness. An autonomous maneuver strategy for dual-UAV olive formation air warfare is provided, which makes use of UAV capabilities such as obstacle avoidance, formation, and confrontation to maximize the effectiveness of the attack.Öğe Optimization strategies for atari game environments: integrating snake optimization algorithm and energy valley optimization in reinforcement learning models(Multidisciplinary Digital Publishing Institute (MDPI), 2024) Sarkhi, Sadeq Mohammed Kadhm; Koyuncu, HakanOne of the biggest problems in gaming AI is related to how we can optimize and adapt a deep reinforcement learning (DRL) model, especially when it is running inside complex, dynamic environments like “PacMan”. The existing research has concentrated more or less on basic DRL approaches though the utilization of advanced optimization methods. This paper tries to fill these gaps by proposing an innovative methodology that combines DRL with high-level metaheuristic optimization methods. The work presented in this paper specifically refactors DRL models on the “PacMan” domain with Energy Serpent Optimizer (ESO) for hyperparameter search. These novel adaptations give a major performance boost to the AI agent, as these are where its adaptability, response time, and efficiency gains start actually showing in the more complex game space. This work innovatively incorporates the metaheuristic optimization algorithm into another field—DRL—for Atari gaming AI. This integration is essential for the improvement of DRL models in general and allows for more efficient and real-time game play. This work delivers a comprehensive empirical study for these algorithms that not only verifies their capabilities in practice but also sets a state of the art through the prism of AI-driven game development. More than simply improving gaming AI, the developments could eventually apply to more sophisticated gaming environments, ongoing improvement of algorithms during execution, real-time adaptation regarding learning, and likely even robotics/autonomous systems. This study further illustrates the necessity for even-handed and conscientious application of AI in gaming—specifically regarding questions of fairness and addiction.Öğe Optimizing MIMO-OFDM communication systems: a comparative analysis of LMMSE and LZF equalizers under varying antenna configurations(Association for Computing Machinery, 2024) Al-Aanbagi, Sarah; Koyuncu, HakanOrthogonal Frequency Division Multiplexing (OFDM) has emerged as a fundamental modulation technique in modern broadband wireless communication systems, effectively addressing challenges arising from dispersive channels and random fading phenomena. By employing orthogonal subcarrier frequencies, OFDM enables the simultaneous transmission of multiple in-formation symbols, thereby mitigating Inter-Symbol Interference (ISI) often encountered in scenarios characterized by multi-path propagation. In contrast to traditional Frequency Division Multiplexing (FDM) methods, OFDM eliminates the need for guard bands between adjacent frequency bands, thereby improving spectral efficiency for high-speed data transmission. This paper conducts an extensive analysis of Zero Forcing (ZF) and Minimum Mean Squared Error (MMSE) equalizers in wireless Multi-Input Multi-Output (MIMO) systems employing BPSK, QPSK, and 16-PSK modulations. Lever-aging MATLAB-based simulations, diverse configurations involving multiple transmit and receive antennas are explored, yielding insightful findings concerning output Signal-to-Noise Ratio (SNR) and Bit Error Rate (BER). Our study underscores the efficacy of OFDM-MIMO systems with equalizers in enhancing communication performance across a spectrum of wireless scenarios, offering valuable insights for both theoretical exploration and practical implementation.Öğe Optimizing multi neural network weights for COVID-19 detection using enhanced artificial ecosystem algorithm(International Information and Engineering Technology Association, 2023) Koyuncu, Hakan; Arab, MunafThe role of machine learning in medical research, particularly in addressing the COVID-19 pandemic, has proven to be significant. The current study delineates the design and refinement of an artificial intelligence (AI) framework tailored to differentiate COVID-19 from Pneumonia utilizing X-ray scans in synergy with textual clinical data. The focal point of this research is the amalgamation of diverse neural networks and the exploration of the impact of metaheuristic algorithms on optimizing these networks' weights. The proposed framework uniquely incorporates a lung segmentation process using a pre-trained ResNet34 model, generating a mask for each lung to mitigate the influence of potential extraneous features. The dataset comprised 579 segmented X-ray images (Anteroposterior and Posteroanterior views) of COVID-19 and Pneumonia patients, supplemented with each patient's textual medical data, including age and gender. An enhancement in accuracy from 94.32% to 97.85% was observed with the implementation of weight optimization in the proposed framework. The efficacy of the model in detecting COVID-19 was further ascertained through a comprehensive comparison with various architectures cited in the existing literature.
