Yazar "Kurnaz, Sefer" seçeneğine göre listele
Listeleniyor 1 - 20 / 91
Sayfa Başına Sonuç
Sıralama seçenekleri
Öğe A high efficiency thyroid disorders prediction system with non-dominated sorting genetic algorithm NSGA-II as a feature selection algorithm(Institute of Electrical and Electronics Engineers Inc., 2020) Kurnaz, Sefer; Mohammed, Mohammed Sami; Mohammed, Sahar JasimIn spite of availability of patient's data in hospitals, health care institute and websites but still hard to collected especially for a risk disease like thyroid disorders. A new model by using Non Sorting Genetic Algorithm are selected for rows reductions and attributes selected with a three data mining techniques for a faster and accurate thyroid disorders detection. Two types of thyroid disorders with 4 different classes for each type are used for this design, in addition 500+972 are used with 29 attributes as training and testing data respectively with cross validation=5. Performances of this model are measured by using some parameter as accuracy , precision , etc. This model is studied for using all/some features with the proposed model and compare it with Sequential model. A scatter plot and area under curve are also presented in this work for training data to show the classes predication enhancement. © 2020 IEEE.Öğe A hybrid CNN-LSTM approach for precision deepfake image detection based on transfer learning(Multidisciplinary Digital Publishing Institute (MDPI), 2024) Al-Dulaimi, Omar Alfarouk Hadi Hasan; Kurnaz, SeferThe detection of deepfake images and videos is a critical concern in social communication due to the widespread utilization of deepfake techniques. The prevalence of these methods poses risks to trust and authenticity across various domains, emphasizing the importance of identifying fake faces for security and preventing socio-political issues. In the digital media era, deep learning outperforms traditional image processing methods in deepfake detection, underscoring its significance. This research introduces an innovative approach for detecting deepfake images by employing transfer learning in a hybrid architecture that combines convolutional neural networks (CNNs) and long short-term memory (LSTM). The hybrid CNN-LSTM model exhibits promise in combating deep fakes by merging the spatial awareness of CNNs with the temporal context understanding of LSTMs. Demonstrating effective performance on open-source datasets like “DFDC” and “Ciplab”, the proposed method achieves an impressive precision of 98.21%, indicating its capability to accurately identify deepfake images with a limited false-positive rate. The model’s error rate is 0.26%, emphasizing the challenges and intricacies inherent in deepfake detection tasks. These findings underscore the potential of hybrid deep learning techniques for addressing the urgent issue of deepfake image detection.Öğe A new localization mechanism in IoT using grasshopper optimization algorithm and DVHOP algorithm(Springer, 2023) Janabi, Shakir Mahmood Al; Kurnaz, SeferNowadays, different types of computer networks such as Wireless sensor networks (WSNs), the Internet of things (IoT), and wireless body area networks (WBANs) transfer information, share resources, and process information. The IoT is a novel network which interconnects various smart devices and can consist of heterogeneous components such as WSNs for monitoring and collecting information. Characterized by specific advantages, the IoT contains different types of nodes, each with few sensors to collect environmental information on agriculture, ecosystem, search and rescue, conflagrations, etc. Despite extensive applications and high flexibility in the modern world, the IoT faces specific challenges, the most important of which include routing, energy consumption and localization. Localization leads to other network challenges and thus can be considered the most important challenge in the IoT. Localization refers to a process aiming at determining the positions and locations of objects lacking global positioning system (GPS) and needing to use the information of network sensors and topology to estimate their own positions and locations. The distance vector hop (DV-Hop) algorithm is a range-free localization technique, in which the major challenge is that the number of hops between two nodes is multiplied by a number that is the same for all nodes leading to a significant reduction in the localization accuracy. In the method proposed in this paper, a network node with no GPS determines the hops from three anchor nodes with GPS. The location of smart objects can be then estimated according to distances from those anchor nodes. Thereafter, a few positions can be created nearby to mitigate the error. Then each position can be regarded as a member of the grasshopper optimization algorithm (GOA) to minimize the localization error. According to the results obtained from implementation of the proposed algorithm, it is characterized by a lower localization error than grasshopper optimization, butterfly optimization, firefly and swarm optimization algorithms.Öğe A novel function of a research process based on a power internet of things architecture intended for smart grid demand schemes(Multidisciplinary Digital Publishing Institute (MDPI), 2024) Al-Mashhadani, Sarmad Waleed Taha; Kurnaz, SeferThe global energy sector is currently undergoing a significant transformation to address sustainability, energy efficiency, and grid resilience. Smart grids, leveraging advanced technologies like the power internet of things (PIoT), play a crucial role in this transformation. This research focuses on enhancing the efficiency, reliability, and sustainability of electricity distribution through IoT technologies. It envisions a system where interconnected devices, sensors, and data analytics optimize energy consumption, monitor grid conditions, and manage demand response scenarios. Central to this effort is the integration of PIoT into the smart grid infrastructure, particularly in implementing dynamic pricing strategies for demand response. Leveraging power line communication (PLC) techniques, this innovative approach facilitates real-time communication between grid components and consumers. The results demonstrate improved grid stability through dynamic load management, effectively responding to demand fluctuations, and minimizing disruptions. The deployment of dynamic pricing methods using PLC-driven schemes empowers customers by offering access to realtime energy use data. This access incentivizes energy-efficient behavior. leading to a 30% increase in the adoption of energy-saving techniques among consumers. A utility company pilot study claimed a 12% dropÖğe A systematic literature survey of emotion detection of students in e-learning(Institute of Electrical and Electronics Engineers Inc., 2022) Ibrahim, Qais Rashid; Kurnaz, SeferAffective computing is a part of artificial intelligence, which is becoming more important and widely used in education to process and analyze large amounts of data. Consequently, the education system has shifted to an E-learning format because of the COVID-19 epidemic. Then, e-learning is becoming more common in higher education, primarily through Massive Open Online Courses (MOOCs). This study reviewed many prior studies on bolstering educational institutions using AI methods, including deep learning, machine learning, and affective computing. According to the findings, these methods had a very high percentage of success. These studies also helped academic institutions, as well as teachers, understand the emotional state of students in an e-learning environment.Öğe Advanced hybrid and preprocessing models for diagnosis challenges in data classification(Engineering and Technology Publishing, 2024) Fayez, Mustafa Adil; Kurnaz, SeferMachine Learning (ML), often viewed as a cutting-edge technology best suited for qualified specialists, presents limited access for other physicians and scientists in the medical profession. In this work, we provide a new, sophisticated, and highly successful technology for medical applications, especially cardiac diagnostics. We propose a novel advanced hybrid optimization model with two essential parts. Initially, we apply a high-performance hybrid resampling technique for feature engineering and pre-processing. This approach, which combines Synthetic Minority Oversampling Technique Edited Nearest Neighbors (SMOTEENN) with Neighborhood Cleaning Rules (NCL), addresses class imbalance in the data. We developed a complex hybrid optimization model that incorporates hyper-parameter optimization, advanced Application Programming Interface (API) functions, and a super-learner ensemble model to enhance diagnosis accuracy in cases where datasets lack balance. Furthermore, we developed high-performance prediction models using sophisticated Support Vector Machines (SVMs). We show that, with re-sampled Cardiovascular Disease (CVD) data, the advanced hybrid optimization model attained an astounding accuracy of 98%. By comparison, an advanced SVM model obtained 96% accuracy, while an advanced deep learning model produced 95.5% accuracy. Our new sophisticated hybrid optimization machine learning models may significantly improve physicians’ interpretation of ML results. This strategy could make it easier to apply AI methods on a large scale in the clinic, which would eventually raise patient outcomes and diagnostic accuracy.Öğe An effective mechanism for FOG computing assisted function based on Trustworthy Forwarding Scheme (IOT)(Multidisciplinary Digital Publishing Institute (MDPI), 2024) Hameed, Fatimah Mohammed Hameed; Kurnaz, SeferAs the Internet of Things (IoT) continues to proliferate, the demand for efficient and secure data processing at the network edge has grown exponentially. Fog computing, a paradigm that extends cloud capabilities to the edge of the network, plays a pivotal role in meeting these requirements. In this context, the reliable and trustworthy forwarding of data is of paramount importance. This paper presents an innovative mechanism designed to ensure the trustworthiness of data forwarding in the context of MQTT (Message Queuing Telemetry Transport), a widely adopted IoT communication protocol. Our proposed mechanism leverages the inherent advantages of MQTT to establish a robust and secure data-forwarding scheme. It integrates fog computing resources seamlessly into the MQTT ecosystem, enhancing data reliability and security. The mechanism employs trust models to evaluate the credibility of IoT devices and fog nodes involved in data forwarding, enabling informed decisions at each stage of the transmission process. Key components of the mechanism include secure communication protocols, authentication mechanisms, and data integrity verification. The proposed secure communication protocols (TLS/SSL, MQTTS, and PKI) and data integrity verification methods (MAC, digital signatures, checksums, and CRC) provide a robust framework for ensuring secure and trustworthy data transmission in IoT systems. These elements collectively contribute to the establishment of a reliable data forwarding pipeline within MQTT. Additionally, the mechanism prioritizes low-latency communication and efficient resource utilization, aligning with the real-time requirements of IoT applications. Through empirical evaluations and simulations, the research demonstrates the effectiveness of our proposed mechanism in improving the trustworthiness of data forwarding, while minimizing overhead, as the experiment was conducted with 15 fog nodes, and the maximum Level of Trust (LoT) score was 0.968, which is very high, with an estimated accuracy of 97.63%. The results indicate that our approach significantly enhances data security and reliability in MQTT-based IoT environments, thereby facilitating the seamless integration of fog computing resources for edge processing.Öğe An improved image steganography security and capacity using ant colony algorithm optimization(Tech Science Press, 2024) Jasim, Zinah Khalid Jasim; Kurnaz, SeferThis advanced paper presents a new approach to improving image steganography using the Ant Colony Optimization (ACO) algorithm. Image steganography, a technique of embedding hidden information in digital photographs, should ideally achieve the dual purposes of maximum data hiding and maintenance of the integrity of the cover media so that it is least suspect. The contemporary methods of steganography are at best a compromise between these two. In this paper, we present our approach, entitled Ant Colony Optimization (ACO)-Least Significant Bit (LSB), which attempts to optimize the capacity in steganographic embedding. The approach makes use of a grayscale cover image to hide the confidential data with an additional bit pair per byte, both for integrity verification and the file checksum of the secret data. This approach encodes confidential information into four pairs of bits and embeds it within uncompressed grayscale images. The ACO algorithm uses adaptive exploration to select some pixels, maximizing the capacity of data embedding while minimizing the degradation of visual quality. Pheromone evaporation is introduced through iterations to avoid stagnation in solution refinement. The levels of pheromone are modified to reinforce successful pixel choices. Experimental results obtained through the ACO-LSB method reveal that it clearly improves image steganography capabilities by providing an increase of up to 30% in the embedding capacity compared with traditional approaches; the average Peak Signal to Noise Ratio (PSNR) is 40.5 dB with a Structural Index Similarity (SSIM) of 0.98. The approach also demonstrates very high resistance to detection, cutting down the rate by 20%. Implemented in MATLAB R2023a, the model was tested against one thousand publicly available grayscale images, thus providing robust evidence of its effectiveness.Öğe An investigational FW-MPM-LSTM approach for face recognition using defective data(Elsevier, 2023) Mahmood, Baraa Adil; Kurnaz, SeferFacial recognition systems are based on the features and traits of the face, since the systems are classified as biometric systems. Additionally, they are founded on the image processing, machine vision and machine learning principles. From images, imperfect information is considered by face recognition systems. A variety of image reconstruction mechanisms is vital in this situation in order to match faces. The proposed method calls for image enhancement at the pre-processing stage. Following the image segmentation and reconstruction stage, the best facial features are extracted using features such the eyes, cheeks, face area and lips. By means of fractal model and wavelet transform the operation is performed. Using the Moore Penrose Matrix, the LSTM neural network is then improved also known as the MPM-LSTM, to train and test the system. From experimental results, the outcomes show that the proposed methodology performs better than the contemporary techniques.Öğe Applying datamining techniques to predict hearing aid type for audiology patients(Inst Information Science, 2020) Aljabery, Maalim A.; Kurnaz, SeferOur research is primarily based on dealing with different types of data using Data Mining (DM) techniques. In this research, we devoted ourselves to determining the type of Hearing Aid (HA) needed by patients with hearing impairment. HA type Diagnosis is a medical application that is a major challenge for researchers. Using DM techniques and Machine Learning (ML) has created a major challenge in the process of predicting the appropriate HA type for Audiology Patients (APs). Thus, this problem is primarily in the domain of classification problems. Our study makes a summary of some technical articles on determining the specific type of HA and introduces a study of using DM techniques to improve the accuracy predict for this purpose. Furthermore, our research includes the creation of a new Audiology Dataset based on the addition of some important fields on the old audiology database and analyses a new data of APs. These data have been obtained from the field work for nearly eight consecutive years, then extract a new classification based on this analysis. Relied on our search to reach the highest degree of accuracy in predicting the type of appropriate HA for APs who use it to enhance their hearing, we applied, compared, and analyzed the Neural Network (NN) and Support Vector Machine (SVM), applying Anaconda Navigator version 1.7.0, Orange Canvas version 3.13.0, and Spyder version 3.2.6 applications for Python coding.Öğe Artık Ağ Tabanlı Uygulamayla Gözlerde Bulunan Bakterilerin Sınıflandırılması(2024) Özçınar, Betül; Kurnaz, SeferAraştırmada, ResNet mimarisi kullanılarak TensorFlow ve Keras kütüphaneleri kullanılarak bir derin öğrenme modeli oluşturulmuştur. Çalışmada 6 farklı bakteri sınıfı için toplamda 689 adet bakteri resmi veri kümesi olarak kullanılmıştır. Yazılım tasarımı, veri ön işleme, model oluşturma ve eğitim adımlarını içermektedir. Veri ön işleme aşamasında, resimler normalize edilmiş ve boyutlandırılmıştır. Model oluşturma aşamasında, ResNet mimarisi tercih edilmiştir çünkü derin ağların daha iyi öğrenme yetenekleri sunabileceği bilinmektedir. Model eğitimi sırasında, eğitim verisi üzerinde iteratif bir yaklaşım benimsenmiş ve optimize edici işlevler kullanılarak ağın ağırlıkları ayarlanmıştır. Sonuçlar, tasarlanan yazılımın %83,33 doğruluk oranı ile bakteri resimlerini başarılı bir şekilde sınıflandırdığını göstermektedir. Bu sonuçlar, derin öğrenme tekniklerinin biyomedikal görüntü analizinde potansiyelini vurgulamaktadır. Bu çalışma, bakteri sınıflandırma konusunda daha geniş veri kümeleri ve daha gelişmiş özellik mühendisliği tekniklerinin entegrasyonunu içerecek şekilde genişletilebilir.Öğe Automatic combination of user views for database creation(Altınbaş Üniversitesi, 2020) Kurnaz, Sefer; Başar, ErdenPeople make up their minds in accordance with the information that they have in hand. If they have the right information the decision will be right otherwise it will be inaccurate. Information can be created on demand, which means when information is needed it should be created as soon as possible. We can ask ourselves “What is the information?” If it is going to help me make my mind, I have to define it. Information can be defined as processed data. Data itself means raw information, while information is the form of meaningful data. In this study we are presenting how to build our data base to save information in correct form to create useful information. (Kurnaz and Başar, 1991)Öğe Binary Error and a three-level division multiplexing Dispersion analysis of a rat optical fibre communication system(Institute of Electrical and Electronics Engineers Inc., 2022) Almukhtar, Ali Hussein Oleiwi; Kurnaz, SeferIn this paper, we look at how well the Three Level Code Division Multiplexing (3LCDM) method works in high-speed optical fibre networking environments. Results reveal that the dispersion tolerance of a 50 Gb/s 3LCDM system is higher than that of a standard 50 Gb/s non return to zero (NRZ- OOK) system. The top level has a chromatic dispersion tolerance of 81 ps/nm at BER of 10- • 50 Gb/s, while the lower level has a tolerance of 98 ps/nm for both positive and negative chromatic dispersions. These numbers are more than the typical NRZ values for 40 Gb/s, which are about 49 ps/nm.Öğe Blockchain and Smart Contracts to Improve Dental Healthcare for Children in Primary School(Institute of Electrical and Electronics Engineers Inc., 2021) Al-Rubaye, Wildan Mohammed Araby; Kurnaz, SeferPrivacy preserving is a matter of great importance in many sensitive fields including financial and healthcare areas. As the blockchain technology has achieved an obvious success in privacy protection at the financial level when the Bitcoin has been introduced in 2008, there are considerable attempts and later on successful projects to introduce the blockchain technology to the healthcare fields, where the patient privacy and access control to the personal information is at high priority. The aim of our paper mainly considers the introduction of the smart contracts that are based on the blockchain technology as a method to preserve the patient privacy and improve the reality of dental healthcare providing services.The main focus will be on the treatment plan of the patient by building a smart contract system between multiple partners. By using 'the five types models' our proposed system has been built combining the advantages of blockchain technology and the benefits of smart contracts. The programming language that we have used is Solidity along with proof of authority algorithm on Ethereum blockchain. The proposed smart contract system which is based on blockchain technology can provide the privacy preservation of patient information in addition to the digitalization of healthcare data. © 2021 IEEE.Öğe Classified VPN network traffic flow using time related to artificial neural network(Tech Science Press, 2024) Mohamed, Saad Abdalla Agaili; Kurnaz, SeferVPNs are vital for safeguarding communication routes in the continually changing cybersecurity world. However, increasing network attack complexity and variety require increasingly advanced algorithms to recognize and categorizeVPNnetwork data.We present a novelVPNnetwork traffic flowclassificationmethod utilizing Artificial Neural Networks (ANN). This paper aims to provide a reliable system that can identify a virtual private network (VPN) traffic fromintrusion attempts, data exfiltration, and denial-of-service assaults.We compile a broad dataset of labeled VPN traffic flows from various apps and usage patterns. Next, we create an ANN architecture that can handle encrypted communication and distinguish benign from dangerous actions. To effectively process and categorize encrypted packets, the neural network model has input, hidden, and output layers. We use advanced feature extraction approaches to improve theANN's classification accuracy by leveraging network traffic's statistical and behavioral properties.We also use cutting-edge optimizationmethods to optimize network characteristics and performance. The suggested ANN-based categorization method is extensively tested and analyzed. Results show the model effectively classifies VPN traffic types.We also show that our ANN-based technique outperforms other approaches in precision, recall, and F1-score with 98.79% accuracy. This study improves VPN security and protects against new cyberthreats. Classifying VPNtraffic flows effectively helps enterprises protect sensitive data, maintain network integrity, and respond quickly to security problems. This study advances network security and lays the groundwork for ANN-based cybersecurity solutions.Öğe Correction to: Deep neural network for human falling prediction using log data from smart watch and smart phone sensors (Soft Computing, (2023), 10.1007/s00500-023-09295-2)(Springer Science and Business Media Deutschland GmbH, 2024) Al-Shawi, Anas Nabeel; Kurnaz, SeferThe co-author names which previously reads Sefer Krnez Should read as: Sefer Kurnaz. The original article has been corrected. © Springer-Verlag GmbH Germany, part of Springer Nature 2024.Öğe CSK based on priority call algorithm for detection and securing platoon from inside attacks(2020) Al-Sheikhly, Mohammed; Kurnaz, SeferThe platooning is an emerging concept in VANETS that involves a group of vehicles behaving as a single unit via the coordination of movement. The emergence of autonomous vehicles has bolstered the evolution of platooning as a trend in mobility and transportation. The autonomous vehicles and the elimination of individual and manual capabilities introduces new risks. The safety of the cargos, passenger and the advanced technology had increased the complication of the security concerns in platooning as it may attract malicious actors. In improving the security of the platoon, the threat and their potential impacts on the vehicular systems should be identified to ensure the development of security features that will secure against the identified risks. In this paper, two critical types of security breaches were identified those are Sybil attack and Delay attacks. Those security attacks can be somewhat disruptive and dangerous to the regular operation of the platoon leading to severe injuries, increased fuel consumption and delay the performance of the network. The research in this paper focuses on design, detection and the mitigation of attacks in a vehicle platoon. priority call algorithm in combination with color-shift keying modulation is used to protect the platoon alleviating the undesirable impacts such as collisions, oscillations and disintegration in the platoon caused by the attacks.Öğe Deep fake image detection based on deep learning using a hybrid CNN-LSTM with machine learning architectures as classifier(Institute of Electrical and Electronics Engineers Inc., 2024) Al-Dulaimi, Omar Alfarouk Hadi Hasan; Kurnaz, SeferOne of the most important and difficult subjects in social communication is detecting deepfake images and videos. Deepfake techniques have developed widely, making this technology quite available and proficient enough so that there is worry about its bad application. Considering this issue, discovering fake faces is very important for ensuring security and preventing sociopolitical issues on a private and general level. Deep learning provides higher performance than typical image processing approaches when it comes to deepfake detection. This work presents construction of an artificial intelligence system, which is capable of detecting deepfake from more than one dataset. This study proposes neural network models based on deep learning using random forest (RF) and support vector machines (SVM) as classifier for deepfake detection. The use of two classifiers (RF) and (SVM) and their combination with a convolutional neural network is the first study of its kind in the field of deepfake detection in images from three open-source datasets (FaceForensics++, FaceAntiSpoofing, and iFakeFaceDB). This proposed method shows an accuracy of 96%, 87% and 52% in iFakeFaceDB, CelebA-Spoof, FaceForensics++ and respectively.Öğe Deep learning based COVID-19 detection via hard voting ensemble method(Springer, 2023) Shareef, Asaad Qasim; Kurnaz, SeferHealthcare systems throughout the world are under a great deal of strain because to the continuing COVID-19 epidemic, making early and precise diagnosis critical for limiting the virus’s propagation and efficiently treating victims. The utilization of medical imaging methods like X-rays can help to speed up the diagnosis procedure. Which can offer valuable insights into the virus’s existence in the lungs. We present a unique ensemble approach to identify COVID-19 using X-ray pictures (X-ray-PIC) in this paper. The suggested approach, based on hard voting, combines the confidence scores of three classic deep learning models: CNN, VGG16, and DenseNet. We also apply transfer learning to enhance performance on small medical image datasets. Experiments indicate that the suggested strategy outperforms current techniques with a 97% accuracy, a 96% precision, a 100% recall, and a 98% F1-score.These results demonstrate the effectiveness of using ensemble approaches and COVID-19 transfer-learning diagnosis using X-ray-PIC, which could greatly aid in early detection and reducing the burden on global health systems.Öğe Deep neural network for human falling prediction using log data from smart watch and smart phone sensors(Springer Science and Business Media Deutschland GmbH, 2023) Al-Shawi, Anas Nabeel; Kurnaz, SeferThe purpose of this research was to conduct a prediction human falling using deep learning and dimensionality reduction techniques in human activity recognition and behavioral prediction using smart watch and smart phone data. The deep learning-based techniques combined with multiple sensor data aim to classify daily activities. Previous work in human falling has focused on using multiple accelerometers placed on different parts of the body, with more recent work focused on sensors embedded in smartphones to classify activities. This research classifies activities from utilizing the data from the following sensors—accelerometer, gyroscope and magnetometer. In addition to comparing these evaluation metrics, a comparison of each network’s confusion matrix, feature importance and multisensory fusion analysis is performed—to evaluate which network best suits the data and successfully classifies the daily activities in question. Another intriguing aim of this research is to compare two data clustering techniques for visualizing the smart watch and smart phone dataset. This research aims to present the best visualization technique by conducting a comparative study on the two-visualization techniques. The result of this research found that all six-machine learning classification algorithms consistently outperformed State-of-the-Art baselines. Deep Neural Network (99.97% accuracy) and MLP (90.55%) accuracy performed excellently on the data, with very little misclassified instances. All six-classification algorithms produced more insightful, predictive results than existing baselines, while DNN successfully clustered and visualized the data. The results show that each algorithm is suited to the smart watch and smart phone dataset, with high performance results achieved throughout. The DNN model does not struggles in distinguishing between the falling activity and running activity with 7% of the activity misclassified. DNN outperforms MLP in this aspect as it misclassifies 3% of the activities between jogging and running. A solution to this would be to place an extra sensor on the thigh to distinguish between both activities. This sensor would lead to detection in a greater acceleration and range of motion in the upper thigh area when the subject is running in comparison to falling.
