Theses supervised by Dr. Öğr. Üyesi Ayça Kurnaz Türkben
31 theses · Altınbaş University
A new IoT security framework using hybrid deep learning techniques
Connecting systems, apps, data management, operations, the Internet of Things builds a network, continuously supports organizations while also opening up new avenues for cyberattacks. IoT security is currently seriously threatened by illicit downloading and virus attacks, which have the potential to compromise private data and harm a company's reputation and finances. Here we describe a hybrid deep learning optimization strategy for detecting and averting assaults in Internet of Things environments. We build a cybersecurity warning system index by first identifying and quantifying pertinent aspects, and then we assess the situation. We employ bio-inspired approaches to increase the efficiency of an Intrusion Detection System (IDS) by lowering the dimensionality of the data and eliminating noisy inputs. One such method that improves IDS effectiveness is the Grey Wolf Optimization (GWO) algorithm, which can identify both typical and anomalous network congestion. By using different pre-processing techniques, we have enhanced the intelligent initialization step and made sure that informative features are there right away. To minimize underlying data characteristics in a large data environment, To find and confirm index components, integrate Whale and Grey Wolf optimization with a deep learning strategy in simulation to prevent attacks. TensorFlow is a deep neural network that uses system software plagiarism detection to classify software that has been copied. Our suggested strategy for assessing cybersecurity threats in IoT offers better classification results than current approaches, as evidenced by experimental data. Therefore, we use Whale and Grey Wolf Optimization (WGWO) in combination with a deep convolutional network for efficient attack avoidance in IoT.
Dünyadaki siber suçları azaltmak için gelişmiş yapay tabanlı siber güvenlik ağı
In the current digital epoch, cybersecurity emerges as a paramount concern, given the relentless advancement of cyber threats that pose formidable challenges to global digital infrastructures. This thesis embarks on a comprehensive exploration, centering on advanced neural network models, particularly emphasizing the Convolutional Neural Network (CNN), to address the imperative need for resilient cyber defense mechanisms. Employing meticulous experimentation and analysis utilizing the 'Cyber Security Indexes' dataset, this study meticulously evaluates the performance of these models across a spectrum of cyber-attack types. The findings illuminate the CNN model's robustness and efficacy, portraying its potential in fortifying cybersecurity measures and countering the evolving landscape of threats. Throughout this exploration, with a focus on achieving a 96% accuracy threshold, the study outlines the implications of these advancements in fostering a more secure digital landscape on a global scale. The comprehensive insights drawn from this research collectively underscore the pivotal role of the CNN model in fortifying cybersecurity defenses, offering a beacon of hope in mitigating the escalating cyber threats prevalent in today's digital milieu.
IoT platformunda sürücü dikkat dağılmasını algılamak için LSTM ile evrimsel sinir ağının çeşitleriyle optimum özellikli ayar modeli
Nowadays, traffic accidents are caused due to the distracted behaviors of drivers that have been noticed with the emergence of smartphones. Due to distracted drivers, more accidents have been reported in recent years. Therefore, there is a need to recognize whether the driver is in a distracted driving state, so essential alerts can be given to the driver to avoid possible safety risks. For supporting safe driving, several approaches for identifying distraction have been suggested based on specific gaze behavior and driving contexts. Thus, in this paper, a new Internet of Things (IoT)-assisted driver distraction detection model is suggested. Initially, the images from IoT devices are gathered for feature tuning. The set of Convolutional Neural Network (CNN) methods like ResNet, LeNet, VGG 16, AlexNet GoogleNet, Inception-ResNet, DenseNet, Xception, and mobilenet are used, in which the best model is selected using Self Adaptive Grass Fibrous Root Optimization (SA-GFRO) algorithm. The optimal feature tuning CNN model processes the input images for obtaining the optimal features. These optimal features are fed into the Long Short-Term Memory (LSTM) for getting the classified distraction behaviors of the drivers. From the validation of the outcomes, the accuracy of the proposed technique is 95.89%. Accordingly, the accuracy of the existing techniques like SMO-LSTM, PSO-LSTM, JA-LSTM, and GFRO-LSTM is attained as 92.62%, 91.08%, 90.99%, and 89.87%, respectively for dataset 1. Thus, the suggested model achieves better classification accuracy while detecting distracted behaviors of drivers and this model can support the drivers to continue with safe driving habits.
Ad hoc ağlarda siber güvenlik tehditlerinin tespitinin değerlendirilmesinde yapay zeka kullanımı
This study explores the use of Artificial Intelligence (AI) to enhance cybersecurity in Mobile Ad-hoc Networks (MANETs), which are
Yapay zeka teknolojilerine dayanan 5G ile nitelikler arasındaki saldırı tespitine yönelik siber güvenlik önlemleri
With the emerging cybersecurity arena, especially in the context of next-generation communication networks like 5G networks, precise identification of Distributed Denial of Service (DDoS) attacks is an urgent challenge. The large dimensionality, high-speed data transmission, and high network diversity typical of 5G environments have a propensity to make conventional machine learning models inadequate. To address these problems, this research suggests a new deep learning approach that uses symmetry to help detect unusual activities in today's fast and complex networks. The system architecture under examination is a Tree Convolutional Neural Network (Tree-CNN) that is particularly capable of understanding hierarchical and symmetrical interdependencies among network traffic, prevalent in 5G communications, as their architecture is distributed and layered. Supporting it is a deep autoencoder module that is employed to be capable of extracting latent symmetrical patterns, noise reduction, and improving the discriminative representation of anomaly behaviour. Model learning performance is further enhanced by the addition of a leader-influenced velocity-based spiral optimization algorithm, a novel metaheuristic, which achieves an effective exploration-exploitation trade-off. The aim here is to optimize Tree-CNN parameters, deep autoencoders, and classification thresholds, which results in improved detection accuracy but at the cost of computational practicality. In the scenario of 5G networks—where there is a sense of urgency in real-time processing and adaptive threat response—the necessity for an accuracy-speed trade-off arises. Performance measurements were performed on three benchmark datasets: UNSW-NB15, CIC-IDS 2017, and CIC-IDS 2018, which represent various traffic patterns as one would see in 5G networks, such as bursty rates of traffic and multi-modal inputs. The novel framework achieved exceptional accuracy levels of 96.02% in UNSW-NB15, 99.99% in CIC-IDS 2017, and 99.96% in CIC-IDS 2018, along with nearly perfect precision and recall rates. These results justify the capability of the system to well-detect threats with very low false negatives and positives. While design imposes a medium computational overhead, this is offset by a dramatic enhancement in detection resilience as well as scalability. What is important here is that the model demonstrates its effectiveness in 5G-supported infrastructures, where high-data-rate streams, edge processing, and latency-prone services require high-performance but guaranteed security mechanisms. The proposed symmetry-aware hybrid detection model not only promotes state-of-the-art future 5G networks. It emphasizes the significance of symmetrical pattern detection, hierarchical feature learning, and adaptive optimization as essential components in the construction of next-generation smart security systems.
Computer network trafic classification based ai techniques
Identifying and categorizing network traffic is the initial step in determining what sorts of applications are traveling via the network. An internet service provider or network operator may better control a network's overall performance by using this method. For the classification of network traffic data, new AI methods based on methods are provided in this thesis. The data obtained from Kaggle was used to train the KNN to classify the dataset into normal and abnormal. The aim of KNN is to learn the features of the normal and abnormal cases of traffic and try to classify them. Then a search algorithm is applied to obtain the best parameters for the KNN. The grid search algorithm was applied to optimize the performance of the KNN and presented remarkable results when compared with previous research.
Image quality enhancement and rain removal on images taken under different rain conditions
The visual quality of images recorded on rainy days, even with high-resolution cameras, is worse, which hampers analytical tasks such as object recognition and categorization. Because of this, picture de-raining has been a hot topic in recent years. An artificial neural network is proposed for the de-raining, contrast enhancement, and feature analysis of a single picture in this study. Deterioration of a picture is judged in comparison to the perfection of a picture. We introduce a deep convolutional neural network (CNN)-based BCET for improving contrast and deep network architecture for removing rain for bettering the image quality. GLCM and DWT feature extractions are used for quality assessments as well. A deep residual network (Res-Net) inspired us to develop a deep detail network that immediately reduces the mapping distance from source to destination, making the learning process easier. The model is trained with a high resolution that decreases background interference and concentrates the model on the structure of rain in photos using previous knowledge of the picture domain to enhance the de-rained output. Contrast enhancement and rain removal have been employed to study image features. The significant power swings in visuals caused by rain diminish the quality and usefulness of outdoor vision systems. Both on images that are made up and images that are real, the presented strategy is better than current best practices in terms of both quality and number.
Ambient intelligence and big data mergeed for wireless student health monitoring system in an IoT
Students' health and welfare depend upon the understanding of various factors related to it. The use of ambient intelligence (AmI) and the Internet of things to assure healthcare monitoring and individualized health care is an efficient and cost-effective method of doing so (IoT). Because of the vast quantity of data collected by sensors in Internet of Things systems, computational issues occur. As a result, we investigate an ambient intelligence-assisted health monitoring system using the Internet of things (IoT) that is used to monitoring student health. This system is heavily reliant on wireless sensor networks (WSNs), which play an important role. Storage of data, an effective exchange of data across the healthcare network and sustained data analysis is handled by the cloud. The proposed work provides the real time alerts with student health information with big data. This paper gives an efficient, low-cost smart health monitoring system for society by combining present technology with smart sensors in the form of wearable devices, mobile/web applications and AI.
Hisse senetlerindeki trendlerindeki dalgalanmaları öngörmekderin öğrenme
Recent studies indicate that variations in the value of the stock market are difficult to anticipate due to the large number of unknowns and variables that affect its value on any given day. This includes the current market conditions, investor opinion toward a certain firm, and political developments. Hurriedly and arbitrarily, the pricing marketplaces are selected when setting the stock price. Due to the fact that it is not rare for the stock market to be dynamic and disorderly (due to several reasons), the stock market's direction is categorized as a random process, with more shifting possibilities in short time frames Therefore, an attempt to carry out a price forecast in the stock market can bring great benefits to investors, by increasing the level of information about the financial market, minimizing exposure to financial risk. In this sense, a computational technique called Artificial Neural Networks (ANNs) can be applied. The Artificial Neural Network (ANN) simulates on computers the functioning of the human brain in a simplified way.
Design and simulation of mobile wimax network using opnet modeler
Transmission weakness signals, low bitrate, low distance with heavy and light services on Orthogonal Frequency Division Multiplexing (OFDM) based lowest throughput and the least response could being a big problem should to solve it, additionally depend on the Quality of Service (QoS) that can classify to heavy and light on this network topology to avoid packet data loss with additional IP network based distributed load on overall entire network and nodes, the design of this network can be divided in three common WiMAX Network each on between them work with different number of Base Station connected indirect by a Ethernet Server than consists of Cloud IP and Router were are contribute to solve different type of services when compared to WiMAX heavy delay 25 sec and light delay 28 sec, the heavy throughput 3500 bitrate and light throughput 6400 bitrate, Traffic Sent on heavy and light for packets, 4045 and 5800 is requires for this type of comparison according to the overall system design can be implement by using OPNET Modeler.The utilize of mobile WiMAX network for huge bitrates is able by the utilize of reliability and scalability for the suggested system additionally, based on the open system interconnected layers there are multiple Quality of Service are applied in this proposed work to enhance the WiMAX network by insert the oriented routers, which generates high services for the end users.
Design and implementation of high data rate is-OWC system based DWDM technique
The use of optical wireless communication for high data rate applications is made possible by the use of license-free and wide-bandwidth access techniques, as well as cost-effective solutions. Inter-Satellite Optical Communication (ISOC), a unique technique targeted at creating transmission between satellites, has just lately been offered to the public at large. The transmission of targeting errors, which generates turbulence in the connection, is the most difficult obstacle in creating an inter-satellite communication. This thesis describes the design of a 32-channel DWDM IS-OWC system that has a greater data rate of 1.28 Tbps and uses the NRZ modulation as a consequence of this. The utilized power is 30 dBm and the extension ratio selected was 30 dB, the selected frequency spacing between channel selected to be 1.58 nm. The suggested system was tested across distances ranging from 1000 to 9000 kilometers and was designed around the criteria of QF and BER by using the most appropriate software of optisystem. The collected results show that the proposed method has a better degree of accuracy. In addition, the examination of alternative modulation formats (NRZ and RZ) will be considered in this work. Thus, employing NRZ is much superior than using RZ in terms of producing a higher quality signal while experiencing fewer bit rate error. Obtaining greater performance systems is shown by the average QF and BER acquired at 7000 km, for example, which were 8.11dBm and 1.62e-16, respectively, indicating higher system performance. The results of eye opening compared between the proposed work and a reference work, and it can be seen that the proposed work achieve much clear eye results indicating a less bit errors of 1 and 0 that has been send over 7000 km, indicating the reliability and the efficiency of the proposed system
İçin robotik süreç otomasyonu (RPA) makine öğrenme tabanlı araştırma üzerinde uygulama
Son zamanlarda, robotik süreç otomasyonu kavramı, üretkenliği artırmak için dijital iş alanında yaygın olarak uygulanmaktadır. Ancak, (makine öğrenimi) tahmine dayalı araştırmalarda bu otomasyon konseptinden yararlanmak kapsamlı bir şekilde araştırılmadı ve test edilmedi. Bu nedenle, bu araştırma, günlüklerinin ve izlerinin öznitelikleri aracılığıyla bulut Kullanıcı Kimliğini tahmin etme vaka incelemesinde Robotik Süreç Otomasyonu'nun (RPA) uygulanmasını inceler. Bu, Yapay Sinir Ağları (YSA) kullanan tahmin araştırması görevlerinin manuel görevlerini taklit etmede otomasyon ilkesini uygulamaya yönelik bir yaklaşım modeli önererek gerçekleştirilir. Buna göre, önerilen yaklaşım modeli araştırma görevlerinden (veri toplama, veri ön işleme, YSA uygulama ve Sonuç doğrulama) oluşur. Daha sonra bu araştırma görevleri, otomasyon yazılımı (Makro-kaydedici) kullanılarak manuel olarak ve otomasyon modeli aracılığıyla test edilir. Sonuçlar, tahmin araştırmasının yürütülmesinde önerilen PRA modelinin kullanılmasının, zaman tüketimini azaltmada ve doğruluğu artırmada önemli bir etkiye sahip olduğunu göstermektedir. Sonuçlar, bilgisayar mühendisliğindeki araştırma topluluğu ve araştırma faaliyetlerinin otomasyon çalışması için ilgi çekici olacaktır.
New medical image recognition system for COVID-19 detection usingconvolutional neural network
In this study new method based convolutional neural network presented for covid-19 detection. The proposed method applied to detect covid-19 in three different datasets. The proposed method show remarkable results when compared with other traditional techniques. The proposed method presented more than 99% accuracy, precision and sensitivity. This mean that the combination of the model are very effective to predicate optimum results. The presented model combined CNN, which is new technique, used in feature extraction problems and extracted high level features from input x-ray images. The CNN trained using ALO which is new optimization algorithm and applied in various optimization problems. The using ALO lead to obtain best weight and basis to present best accuracy and detection rate. On the other hand, the genetic algorithm applied for select best features from features that are extracted by CNN. The genetic algorithm role is to minimize the size of CNN output features to decrease the execution time and increase the performance of the classifiers that applied in this study.
Smart grid voltage control using AI based controllers
Electrical micro-grids are, as the name suggests, small electrical networks compared to conventional ones. They encompass the generation, distribution, and consumption of electricity in a small area. These new electrical system structures integrate advanced devices for the production and conversion of electrical energy, and incorporate modern and sophisticated control, automation, and communication strategies In the context of this research, the microgrid utilizes many power-generating sources but lacks a centralized controller or other kind of communication system. A controller is required to regulate the actual power output and reactive power output of several generations simultaneously. Both the actual and reactive power generated by these two diesel generators are under control. Consequently, the resilience of the Microgrid will be enhanced. In the case of a power outage, diesel generators will be utilized to keep the lights on, while alternative energy sources will be used to charge the batteries. Due to the absence of a clear link between the two generators, neural networks are an ideal tool for addressing non-linear problems.
Intellignce face recognition system
The purpose of this research is to utilize the libraries and models of the luxand platforms for the seek of facial recgnitions which pass through four main stages and other many secondary process to achieve the goal of recognition. The system capture the face image and start the process of analyess and face spotting to then for the stage of alignment where is the face features are detected then moving to the extraction stage where are calculation are done , all these calculations and numbers move to the fourth matching process where results are compared with previous stored information for the targeted face , all those processes utilize the locality preserving projections (LPP) .the PCA Linear Discrimination Analysis is used to solely structure the face spaces. The PCA, LDA, and LPP are classified of one of a kind diagrmaed model. The research is going to pass through the Luxaund technique.
Makine öğrenmeyi kullanarak IoT sistemi hata testi (lıneer regresyon)
In the age of the Internet of Things (IoT), day-to-day objects are outfitted with sensors and actuators. It is also referred to as IoT units that gather information and function moves through speaking wi.th every other. Electrical units with constrained sources may additionally be subjected to big stresses or exterior influences that may additionally lead to their failure. In provider environments security is a imperative requirement, failure ought to be averted earlier than it motives any damage. By imposing predictive upkeep (PdM) the use of laptop gaining knowledge of (ML), computer getting to know algorithms can be utilized to predict screw ups and grant ample renovation time. IoT gadgets gather statistics that can be used to educate desktop gaining knowledge of algorithms to apprehend failure patterns of character devices. This thesis introduces (Decision Tree, Logistic Regression, k-nearest neighbors, Random Forest, and Gradient Boosting) algorithms and its addendum to predict the failure of electrical gadgets by means of the usage of Internet of Things devices.
Smart card security authentications using homomorphic encryption
Nowadays, there are a lot of different systems and platforms, and they all want to be as efficient as possible so that they can provide their customers better service. That persons or computers can trust the data they receive and that it has not been changed is the goal of maintaining communication integrity and authenticity. This may be accomplished by ensuring that the data has not been altered. Utilizing cryptographic methods that are considered standard is necessary in order to maintain integrity. In order to ensure the system's safety, a variety of cryptographic procedures are required. Information that is supposed to be kept private might end up being leaked elsewhere. In this article, we will cover the importance of homomorphic encryption, as well as its uses, applications, and limitations. Theoretical assessments of HE systems including the DGHV, Paillier, BGV, and FHEW are carried out, and an introduction to PKI and encryption approaches based on PKI is also provided. When it comes to the usage of both CPU and memory, the algorithms known as Paillier and RSA, which are used for public-key cryptography, are the most essential. We are able to evaluate the performance of the HE algorithms that we have selected by first installing and then executing the corresponding library software. During the course of the testing method, the RSA public-key encryption library is opened as well as installed in order to evaluate the impact that this library has on the amount of CPU and memory that is used. After that, we will discuss future work that will be triggered by our in-depth investigation of the issue, as well as new ideas for enhancing higher education that will arise from our findings as a direct consequence of our investigations.
Survey of cyber security issues in internet of things (IoT)
The Internet of Things (IoT) has revolutionized the way we live, work, and communicate. With billions of connected devices communicating with each other. However, the widespread use of IoT devices has also created new challenges in the area of cyber security. Cybersecurity threats pose a serious risk to the IoT, which can compromise the privacy, safety, and security of users, organizations, and governments. The vast amount of sensitive data transmitted over IoT networks can make them a target for attacks and cyber criminals. Additionally, IoT devices often lack the security measures required to prevent unauthorized access, data theft, and other cyber-attacks. As a result, the rapid growth of IoT devices has created an urgent need for organizations to implement robust cyber security measures that can prevent and mitigate these threats. This requires an integrated and coordinated approach involving government, industry, and academia to develop and implement effective cybersecurity solutions and standards that can protect the IoT and its users. This thesis examined the architecture supporting Internet of Things devices, data aggregation facilities, and communication routes. It then described device, network, and application assaults against a system. Trusted Execution Environment and Trusted Platform Module, the two most common IoT solutions, will be discussed in the second part. It described each approach's main characteristics, working principles, and drawbacks. This shows that solutions are successful but need to be modified over time. The latter section used artificial neural networks to forecast DoS attacks. This thesis illustrates that IoT security may be improved simply. The norm should ultimately standardize Internet of Things device safeguards.
Uzun mesafeli geniş alan ağı (LoRaWAN) ve uygulamaları akıllı çiftlikler ve ormanların izlenmesi
Fires in forests and agricultural systems have devastating impacts and adversely affect human and animal lives, the economy, and the environment. The recent growth of IoT usecases in a wide array of industrial, utility and environmental applications has necessitated the need for connectivity solutions with diverse requirements. We study the fundamentals and design principles of LPWAN technologies pertain to wireless networks tailored for lowpower gadgets that demand broad coverage over a considerable distance, with LoRaWAN serving as a prime example."LPWAN technologies in both licensed and unlicensed bands have been considered to provide connectivity to a high density of devices over larger coverage areas. We design, implement, and test a fire detection system for forests and farms based on LoRaWAN technology. The proposed system offers more accurate fire detection capability in terms of determining actual fire presence and determining the position of fire. Also, the proposed design can detect fires under a higher number of possible conditions compared to other designs.
Görüntü steganografisini geliştirmek için AI sistemi önerin
Steganography is one of the methods used for the hidden exchange of information and it art of invisible communication. The confidential communication is concealed so that it cannot be seen by human senses. Steganography and encryption offer an effective means of achieving that privacy. Conventional methods of image steganography tend to start with the first pixel and proceed with subsequent pixels until the last piece of the hidden message is embedded. In order to increase the security of the steganography system, modern trends involve arbitrarily hiding data in images using various clever algorithms. Numerous methods will be used in this thesis to improve image steganography and provide high levels of security systems and exchange the secret information in secure way. By arbitrarily concealing encrypted data in the cover, a proposed image steganography method based on Ant Colony Optimization (ACO) aims to improve geographic image steganography. Data Encryption Standard (DES) is used to encode a secret communication to increase the security of the suggested system. The use DES algorithm With using breadth algorithm to generate the key to increase the complexity against the attacker and difficult way to detect the original message. Cover image separating into groups (n*n) optimum pixels can discovered through the connection between pixels. Each block has the Ant Colony System (ACS) applied to it in order to determine the texture of colour among the pixels in the block and determine which pixel is best to use to embed the encrypted secret message bits. This process is repeated for each subsequent block until the secret message finished. The testing study demonstrates that in terms of quality (MSE and PSNR) and security. The final experimental results between security techniques (steganography and cryptography) and artificial intelligence (ACO (ACS) with security techniques) explains. Ant Colony System (ACS) of the ACO algorithm outperforms R(RGB-LSB). The PSNR in the R(RGB-LSB) technique is 71 and ACO is 83, according to the findings of the contrast between the ACO and R(RGB-LSB).
Detection of SQL injection attacks
Web applications that employ SQL datasets are at serious risk from SQL injection attacks. Using two separate datasets, this study examined the utility of ML classification approaches for identifying SQL injection threats. On two datasets—one taken from the actual world and the other created artificially—random forest (RF), descion tree (DT), k-nearest neighbours (KNN), gradient boost (GB), xgb classifier, linear SVM (Support Vector Machines), and RBF (Radial Basis Function) SVM were trained. The outcomes demonstrated that both algorithms could precisely and precisely detect SQL injection attacks. The model that was trained using the synthetic dataset outperformed the model that was trained using the real-world dataset. These findings highlight the potential of ML classification for identifying SQL injection attacks as well as the value of employing a variety of datasets to increase model accuracy (ACC).
Intelligent symptoms checker using artificial intelligence
Globally, there is a substantial unmet need for reliable detection of a broad variety of diseases. When trying to establish an early diagnostic tool and a successful therapy, the complexity of the several disease processes and underlying symptoms reported by the patient group offers great difficulties. With the use of machine learning (ML), a subfield of artificial intelligence, researchers, clinicians, and patients can overcome some of these difficulties (AI). Based on recent research, this article discusses how machine learning (ML) is now being utilized to aid in the early detection of a variety of illnesses.
Advancing electronic commerce using data minnig benefits
Although data mining is now being used in a wide variety of contexts, it has traditionally been put to use in the evaluation of big data sets. Recently, several data mining strategies have been suggested and deployed in the more constrained context of online retail. The e-commerce sector makes use of data mining to examine many kinds of data, such as purchases made by customers, the information logged from websites, and even social media activity. Understanding customer behavior, spotting trends, and fine-tuning advertising initiatives all rely on this data. This paper presents the three primary algorithms used in data mining (D.M.) for online business: association, clustering, and prediction. It highlights several advantages of D.M. to e-commerce businesses, including data pre-treatment, pattern mining for sales, and market pattern analysis, all of which may be accomplished with the help of the three data mining algorithms. Moreover, it also investigates the three data mining algorithms that may aid e-commerce firms with tasks like product planning, sales forecasting, basket analysis, CRM, and market segmentation. This research primarily aims to categorize a product into the four categories of Electronics, Household, Books, and Clothing & Accessories and check the accuracy of this classification.
Neural network training method for classifications of diabetic retinopathy image data on matlab
Deep learning was used in this research to develop a method that could identify hard exudates in DR fundus pictures. Patients with diabetic retinopathy need to be able to differentiate between hard exudates and other symptoms of sickness. This innovative technique accurately identifies hard exudates 99.7% of the time. In the future, detection will also include blood flow and microaneurysms in addition to soft exudates. In addition to this, we need to quantify DR by utilizing segmented pictures. There are two approaches that improve model precision. Then, change the dimensions of the picture patch. The next thing that we could do is investigate the role that picture patches have in the accuracy of the forecast. Convolutional neural networks are utilized in the second technique, which performs an analysis on the first and last 16 unpredicted pixels in each row and column.