401
Archived Theses
0
DOIs Assigned
0%
DOI Rate
Discipline
Örnek bir güneş evi prototipi için tasarım, kontrol ve güç izleme çalışması
KÖMÜR, Salim. Örnek Bir Güneş Evi Prototipi İçin Tasarım, Kontrol ve Güç İzleme Çalışması, Yüksek Lisans Tezi, Ankara, 2016. Müstakil evlerde ve apartmanlarda güneş enerjisi kullanımının önünde duran en önemli üç problem; yüksek maliyet, düşük verim ve binaların çatılarındaki fiziki uygunsuzluk olarak karşımıza çıkmaktadır. Bu tez çalışmasında bahsedilen problemlerin her birisine yönelik birer yöntem ele alınmış ve bir proje kapsamında gerekli tasarım, prototip, kontrol ve güç izleme işlemleri gerçekleştirilerek, aşağıda sunulan yöntemlerin uygulanabilirliği incelenmiştir. 1. Yöntem (Kurulum Maliyetinin düşürülmesine yönelik): Direk Doğru akım kullanımı ile invertör masraflarının düşürülmesi. 2. Yöntem (Verimin arttırılmasına yönelik): Güneş takip sistemi kullanılarak verimin arttırılması. 3. Yöntem (Bina çatılarındaki fiziki uygunsuzluk): Güneş takip sistemlerinin özel tasarımlarla çatılarla bütünleşik olarak tasarlanması. Sunulan tez çalışmasında; özgün tasarım çalışmaları ile yenilikçi döner çatı sisteminin binalar ile bütünleşik prototipi geliştirilmiş, geliştirilen kontrol algoritması ile yenilikçi döner çatı sistemi üzerinde yer alan PV panellerin güneşi iki eksende takip etmesi sağlanmış, yenilikçi döner çatı sistem ile yapılan güç izleme çalışmalarında çatı üzeri optimum sabit açılı PV sistemlere göre %15 elektrik üretimi verim artışı ölçülmüştür.
Human identification using palm print images
Palm recognition is one of the research areas which is considered, in recent years. In this thesis, we introduce a new method for human palmprint identification with local binary pattern and Co-occurrence matrix. First, the palm images are preprocessed with morphological technics. Then feature extraction is applied for images. We used local binary pattern (LBP) and gray level Co-occurance matrix (GLCM) for desired features. This approach is tested for 20 people and there are 1, 2, 3 and 4 images from each people. Our method is compared with PCA method. The result shows that proposed method have high accuracy and good performance (%92) for palmprint recognition.
Kurumsal chatbot sistemleri için otomatik cevap üretme
In this work, we have studied the available intent classification and response selection methods for designing a customer service chatbot in Turkish language. We have compared the available chatbot types in the literature. As our main focus is the methods that are easy to adapt and implement, we decided to work on closed domain, task oriented, retrieval based type chatbot, as it suited best to our application. We have compared two implementation alternatives, i.e. intent classification and response selection. Moreover, the effect of including the history of the dialog to the training is experimented. We have suggested a classification method for labeled datasets using Natural Language Inference. As Turkish is an agglutinative language, most of the available methods in the literature does not perform as good as with languages like English. Therefore, we have conducted experiments with different neural network models to observe and compare their performances on various datasets. We compared the performance of state of the arts methods and analyzed the performance of pretrained language models on classification and natural language inference tasks. We have studied the factors that makes the performance of Turkish datasets lower than English datasets and defined basic problems. Then we tried to improve the performance of available methods on Turkish dataset by suggesting solutions to these problems. We have used many Natural Language Processing methods in our experiments. For tokenization specifically, we compared recent methods in the literature and applied to Turkish language. We experimented the effect of these methods on both Turkish and English datasets. In the SWDA dataset, we have obtained 75.22% classification accuracy on test set and introduced new state of the art. In the XNLI corpus, we have obtained 86.35% NLI accuracy on English test set and 79.85% on the Turkish test set.
MIMO for 5th generation heterogeneous communication networks: Modeling and performance analysis
The day-to-day increasing demand of high data rate and connectivity across all of the devices has lead us to the verge of remanipulating the current wireless deployment for the future generation networks. Current 4G deployment either uses small heterogeneous cellular networks(HetNets) or multi-antenna approach to fulfill the demand. Amalgamate of these two features can lead to a significant improvement. This piece of work is a contribution to the analysis of this combined network paradigm. This dissertation proposes a tractable analysis for use of MIMO in HetNets. A downlink analysis by deploying K-tier BSs, with different transmit power and geographical density across all k-tiers is studied. Along with it, localization of different class of BSs is carried out through well behaved independent Point Process, namely Poisson Point Process(PPP). All of this baseline model setup allow the derivation of analytical expression for coverage probability, average rate, signal to interference and noise ratio(SINR) and area spectral efficiency. Insightful results obtained shows that increasing the BSs does not necessarily increase the coverage probability, but it increases by adding less loaded small BSs. For the MIMO based HetNets with a fair spatial model, a tractable framework shows that spreading the many small SISO systems is preferable over fewer MIMO BSs. But with large number of independent SISO systems, it is more preferable to use MIMO BSs with same number of antennas.
Integrative clustering approaches for cancer subtype discovery
Cancer is a heterogeneous disease and identification of cancer subtypes is critical for personalized treatment and drug development. Recently, cancer genome projects have produced multiple types of high-throughput data for thousands of cancer patients. Exploiting the complementary information between different data types can improve finding subtypes. In the first part of this thesis, we apply multi-view kernel k-means to integrate multiple genomic datasets (i.e.,gene expression, DNA methylation and miRNA expression) on two cancer datasets. We show that combining kernels (i.e., that correspond to different views) with learned weights give better clusters compared to combining the kernels uniformly or using each data set independently. We also demonstrate an improved performance compared to existing models that integrate the same data types. In terms of biological significance, Kaplan-Meier analysis shows that our discovered clusters have distinct survival profiles with statistically significant log-rank test p-values. In the second part, we target the high dimensionality problem of genomic datasets by extending a single-view sparse k-means framework to multi-view setting. This extension allows us to perform integrative clustering and feature selection simultaneously by learning both feature weights and view weights. We confirm that performing feature selection improves the clusters for the majority of the datasets. Altogether, our results indicate that integration of multiple genomic characterizations and the application of feature selection enable the discovery of subtypes that improve over current patient stratifications.
A comparative study of kalman filter and its variants in acoustic echo cancellation
In telecommunication, when a far-end speaker's voice is reflected in the near-end, this reflection causes communication problems and it is known as acoustic echo. Despite its wide coverage in the literature, acoustic echo cancellation (AEC) has been mostly examined for linear systems. The number of approaches to mitigate echo in non-linear systems is not sufficient. In this thesis, Kalman filter and its variants such as Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF) and Cubature Kalman Filter (CKF) have been compared for this purpose.
Some machine learning techniques for medical diagnosis
There is a huge increase in the amount of biological data obtained with the great breakthroughs in technology, and the amount of this data increases exponentially every day. The use of data mining algorithms and machine learning methods has become widespread in the field of health, because analysis and interpretation of biological data is very difficult with traditional methods. In this study, biochemistry and hormone values are used to make some prediction on medical diagnosis with some machine learning methods such as linear regression, generalized linear regression, deep learning, random forests, gradient boosted trees and support vector machines. With the analysis of the laboratory results according to the reference ranges, it is tried to reveal some new relations of biochemistry and hormone parameters. Finally, the results are presented and the performance of the model is evaluated.
Deep learning approaches for MRI image classifications
Although there is no treatment for Alzheimer's disease (AD), but an accurate early diagnosis is very important for both the patient and public care. Our proposed model consists of two main experiments traditional machine learning and deep learning approaches. These two studies were carried out using the Alzheimer's disease Neuroimaging Initiative (ADNI) dataset. In our first experiment, we tested different handcrafted feature descriptors to diagnose Alzheimer's disease. But the most powerful and efficient descriptor is Monogenic Binary Coding (MBC) that gives maximum accuracy of 90.14 percent. In the second experiment, we propose a novel deep convolutional neural network for the diagnosis of Alzheimer's disease and its stages using magnetic resonance imaging (MRI) scans. In this research, we propose a novel 3-way classifier to discriminate patients having AD, mild cognitive impairment (MCI), and normal control (NC) using a pre-trained CNN architecture Inceptionv3. The proposed method work on proficient technique to use transfer learning to identify the images by fine-tuning a pre-trained CNN architecture, Inceptionv3. The performance of the proposed system is calculated over the ADNI dataset. The proposed model showed novel results by giving the best overall accuracy of 95.71% for multiclass classification problems. Our experimental results show that the performance of the deep learning approaches is comparatively higher than the traditional machine learning.
Breast cancer classification using deep neural network
Over the last few decades, cases of breast cancer have increased enormously. It is the second most popular cause of deaths in women in both developed and undeveloped countries. 8 out of 100 women face this popular and dangerous disease in their life period. The only way to cure this disease is to detect breast cancer at early stages. Delay in identifying breast cancer leads to an increase in the death rate. An appropriate data representation determines the performance of classification systems. In this work, we have done some classification on Breast Cancer histopathological images from publically available Break-His dataset using machine learning and deep learning techniques. We proposed different classifiers in our work, Resnet-101, Resnet-18, and Densenet-201, etc as Cnn and multiple handcrafted features like LBP, HOG, and MPT for more accurate classification of breast cancer images. Deep learning extract and organizes features from data. We have organized and prepared a competitive comparison of these different implementations by evaluating their accuracy using deep learning and machine learning techniques. We also organized competitive results for handcrafted feature extractors and matched CNN and handcrafted features extractor accuracies with recent work done. We have achieved some enormous results using these different techniques. We brought in light that how deep networks and CNN are taking the place of handcrafted feature extractors in different image classifications.
Computer aided lung cancer diagnoses with deep learning algorithms
Cancer is one of most deadliest diseases in the world right now, and if not diagnosed in early stages, it leads to death of the patients, which means diagnosis of cancer means life or death for precious human lives, one of the leading cancer types is lung cancer, and diagnosis of lung cancer solely depends on the education and experience of individual doctors and radiologists in current world. In order to overcome this limitation as it is directly related to the life of patient, a computerized lung cancer diagnosis algorithm is to be made which will have experience of hundreds of doctors and radiologists to refer from hence increasing the probability of better diagnosis and save countless lives. This project goes through past works done on this topic, tests numerous deep learning and hard crafted machine learning algorithms in order beat past highest accuracies and achieve the highest accuracy possible, the dataset used in this project is the LIDC dataset which is publically available for research purposes on the internet.
Ranking cancer drivers via betweenness-based outlier detection and random walks
Recent cancer genomic studies have generated detailed molecular data on a large number of cancer patients. A key remaining problem in cancer genomics is the identification of driver genes. We propose BetweenNet, a computational approach that integrates genomic data with a protein-protein interaction network to identify cancer driver genes. BetweenNet utilizes a measure based on betweenness centrality on patient specific networks to identify the so-called {\it outlier genes} that correspond to dysregulated genes for each patient. Setting up the relationship between the mutated genes and the outliers through a bipartite graph, it employs a random-walk process on the graph, which provides the final prioritization of the mutated genes. We compare BetweenNet against state-of-the art cancer gene prioritization methods on lung, breast, and pan-cancer datasets. Our evaluations show that BetweenNet is better at recovering known cancer genes based on multiple reference databases. Additionally, we show that the Gene Ontology terms and the reference pathways enriched in BetweenNet ranked genes and those that are enriched in known cancer genes overlap significantly when compared to the overlaps achieved by the rankings of the alternative methods.
A piezoelectric energy harvester for power generation around waistline
This thesis reports energy harvesting from human waistline by using piezoelectric components. A normal person walks 1.5 to 2 miles a day and breathes all day long. These activities can be useful to generate electrical power. Energy harvesting from human body has been investigated previously, but harvesting energy around waistline has been neglected due to the rather low characteristic frequencies of human waistline that makes it difficult to extract power. To the best of our knowledge, this is the first study that utilizes a piezoelectric design to perform energy harvesting from waistline. The piezoelectric components are well integrated within the belt in such a way that they are not physically and cosmetically discomforting. One can wear it and harvest energy throughout a day. Simulations were performed on COMSOL to reveal maximum performance during breathing. The piezoelectric components were inserted in a waist belt worn by a subject, and were characterized for voltage and power generation. During breathing, the maximum open circuit voltage and rms power were measured to be 1.6 V and 1.90 µW, respectively. These values were observed to increase up to 8.60 V and 1.98 µW during walking and 20.2 V and 14.65 µW during running activity. A full wave bridge rectifier was used to convert AC power to rectified DC power, then a rechargeable 1.2V NiMH battery was connected to the rectifier to store the electrical power generated by the harvester. The power of the battery was then used to operate a temperature sensor. While operating the temperature sensor, the piezoelectric energy harvester continuously recharges the battery. This is an important step forward to eliminate the need for battery replacement for future wearable electronics. Keywords: Energy harvesting, power generation, piezoelectric transducers, self-powered, human gait, breathing, body motion, waistline belt
Meteorological drought forecasting using decision tree
Drought is a natural phenomenon that directly affects human life in a region. Therefore, drought classification and prediction models are essential tools for the mitigation of adverse consequences of drought. Beginning with a brief review of different machine learning techniques to predict drought, the applicability of decision tree for drought classification and the prediction was investigated in this thesis. For this aim, quantitative values of Standardized Precipitation Evapotranspiration Index (SPEI) were collected from a global SPEI data repository (grid points across Antalya province) during the period of 1961-2015. A dataset including 659 samples was created and considered as features in this study. The target SPEI values were first calculated by averaging SPEI values of nearby grid points. Then, drought classes for target SPEI values were determined in terms of wet and dry events. It is well documented that negative monthly SPEI values in a region generally represent dry months at the given region. The lower the SPEI, severe the drought. A binary classification decision tree model was presented that uses the SPEI values of each grid point at month t as inputs and predicts the drought classes (wet or dry) of Antalya at one month later (i.e., month t+1). The dataset was divided into train and validation sets. The predicted and observed values of drought classes were compared using performance evaluation criteria such as total accuracy, kappa statistics, and classification error. It was obtained that a value of 80.63, 19.37, and 0.61 for total accuracy, classification error, and kappa statistics, respectively. This indicated that the presented decision tree model is moderately accurate for drought prediction in the case study region. On the other hand, results showed an increasing trend for dry events in the most recent years that can be considered as the signal for climate variability in the study area.
Theoretical analysis and simulations on the development of an ultrasonic biosensor for the detection of arterial rupture
An aneurysm is a disease related to an artery's distention, which is caused by prolonged high blood pressures and the high elasticity in the artery wall. There are usually no symptoms associated with this disease. However, a long-term aneurysm can be fatal and dangerous, which leads to life-threatening internal bleeding. When an aneurysm is detected in a patient, the best approach to counter this disease is by adding a stent to an artery. The stent insertion may be rather straight-forward for young patients but may pose vital risks for older people. Therefore, it may be best to wait for the surgery until the artery starts rupturing. The waiting period depends on the patient's condition and may vary for each patient from months to several years. When the artery starts rupturing, the stent procedure should be applied since the vital risk is unavoidable. This research has developed a sensor design that can detect this last possible moment when the artery starts rupturing. This sensor will delay the risks related to a coronary artery stent's insertion, giving the maximum possible time to prevent the surgical procedure. When this sensor is fully implemented, the artery rupture can be detected early, and mortality can be prevented, providing a longer, surgery-free, and comfortable life to a patient. This device is envisioned to be placed on the skin aligned with the aneurysm location. The signals received and processed by this sensor will be different for each rupture size, which will provide an idea about rupture size so that the doctor can proceed according to the situations mentioned above. The sensor design was developed on the COMSOL and simulated for different frequency values to detect the artery rupture. A significant difference between the frequency spectrums of the normal and ruptured artery's ultrasonic echoes was observed. Based on the results, a piezoelectric transducer geometry was designed to send and receive ultrasonic signals at an MHz range frequency. Necessary fabrication procedures to manufacture the device were determined. Keyword: Biosensors, ultrasonics, Aneurysm, Arterial, Rupture
Alt taşıyıcı güç modülasyonu ile ortogonal frekans bölmesi çoğaltma performansını iyileştirmek için STBC ve geri yayılma yapay sinir ağı
Alt Taşıyıcı-Güç Modülasyonu (OFDM-SPM) ile Ortogonal Frekans Bölmeli Çoğullama, son zamanlarda umut verici olarak önerilmiştir, OFDM-SPM tekniği, çoklu giriş-tek çıkışlı (MISO) Alamouti uzay-zaman blok kodlaması (STBC) ile birleştirilmiştir. ) çok yollu bir Rayleigh sönümleme kanalı üzerinden kablosuz sistemin bunların kombinasyonlarının performansını incelemek, araştırmak ve ölçmek için kurulmuştur. STBC'nin OFDM-SPM ile iyi çalışmasını sağlamak için, SPM güç bitlerini algılamak için uygun bir eşitleyici önerilmiş ve BER'de önemli bir gelişme olduğu gözlemlenmiştir. Alt taşıyıcı optimize edilmiş güç yeniden atama şemasının, önerilen iletim şeması için en iyi BER performansını sağladığı bulundu. Buna karşılık, SPM için PSP, ortalama bir BER'de orta düzeyde bir gelişme sağladı. Daha yüksek veri hızlarına ulaşmak için performans ölçütlerinde daha fazla iyileştirmeye ihtiyaç vardır. Alıcı tarafında İleri Beslemeli Geri Yayılımlı Yapay Sinir Ağı (BPANN) uygulayarak BER ve Verimi daha da geliştirdik. OFDM-SPM-STBC ile BPANN uygulayarak BER'de ortalama 7 dB SNR kazancı gözlemledik.
Gameplay analysis with graph theory: a sid meier's civilization inspired case study
Traditional games with entertainment purposes have been used as potential experimental areas for research activities which include but not limited to mathematical analysis, artificial intelligence and human behavior. The computational capabilities of modern computers have increased the complexities of these games and eventually formed the generation of video games. This study takes the turn-based strategy video game franchise Sid Meier's Civilization as a sample case for mathematical analysis and gameplay evaluation. Specifically, the suitability of graph theory in the evaluation of the game has been explained along with a conceptually developed game to resemble the Civilization's core gameplay.
Novel waveform designs for future wireless systems:non-coherent orthogonal frequency division multiplexing with subcarrier power modulation (NC-OFDM-SPM)&multi-user auxiliary signal superposition transmission (MU-AS-ST)
Future wireless systems are expected to serve very challenging requirements such as enhancing the spectral efficiency and transmission reliability, securing the transmission and guaranteeing low-complexity and low latency communications. In this scope, we investigate and propose in this work some promising research directions for ensuring an effective design for future wireless systems: 1) we study the combination of multi-dimensional OFDM modulations and non-coherent detection for enhancing the spectral efficiency and reducing the complexity in the design of future wireless systems. Particularly, in this regard, we propose and study a new technique termed as 'Non-Coherent Orthogonal Frequency Division Multiplexing with Subcarrier Power Modulation (NC-OFDM-SPM)' for doubling the spectral efficiency per receiving user/device through the exploration of the power of the subcarriers inside an OFDM block as an additional dimension for conveying extra information. The use of non-coherent detection ensures low-design complexity in this idea. 2) We propose a novel physical layer security design for effective and secure future multiple access communications. The proposed design is called 'Multi-User Auxiliary Signal Superposition Transmission (MU-AS-ST)' which is presented as an alternative design for the current conventional Power Domain NOMA which was studied by the 3GGP (3rd Generation Partnership Project) from release 13 till 16 under the name 'Multi- User Superposition Transmission (MUST)' before being eliminated from the study items in release 17. The proposed design superimposes auxiliary signals with the users data for can- celling the inter-user interference fully while achieving perfect secrecy against both internal (presence of an untrusted legitimate user) and external eavesdroppers. MU-AS-ST achieves better reliability than conventional NOMA and does not use Successive Interference Cancellation. Moreover, this design works for the combination of any two users regardless of their distance from the base station unlike conventional NOMA which works only for the cases where there exists a significant path-loss channel difference between paired (or super-imposed) users. Furthermore, carrying all the processing at the base station makes this design an appealing choice for processing-restricted communication devices such as IoT devices. 3) We study the integration of multi-dimensional OFDM modulation formats in multiple access setups for enhancing the spectral efficiency per area and per device for a more optimal usage of the spectrum allocated for wireless communications. As an example of this integration, we study the combination of OFDM-SPM with MU-AS-ST where we show that this leads to doubling the spectral efficiency per area and per device. The proposed designs were studied thoroughly and their performance was evaluated in terms of different performance metrics such as bit error rate, spectral efficiency, design complexity and peak to average power ratio (PAPR). Keywords: auxiliary signals, eavesdropping, IoT, low-complexity, multi-dimensional OFDM, multiple access, non-coherent detection, physical layer security, power domain NOMA, reliability, spectral efficiency, subcarrier power modulation, wireless communications.
Multiple input multiple output with antenna number modulation and its applications
Multi-user applications of multiple input multiple output systems with antenna number modulation (MIMO-ANM) are proposed in this thesis to be used instead of conventional MIMO and its other variations for providing more advantageous and reliable data transmission in future wireless systems. In the MIMO-ANM scheme, the additional data bits are conveyed by changing the number of active transmit antennas instead of their indices. To utilize all available antennas of MIMO-ANM, Multi-User MIMO-ANM and Multi-User OFDM-SNM are proposed in this thesis as novel multi-access data transmission methods that target serving multiple users by dedicating the antennas/subcarriers used for implementing ANM/SNM to serve a far user, whereas the remaining antennas/subcarriers are used to send data for a near user. The acquired results show that MIMO-ANM and its applications for different communication scenarios offer highly competitive performances and desirable benefits for future wireless networks.
Ionic current estimations from cardiac action potentials by using single and multi-output regression models
Conventional experimental procedures for capturing the dynamics of ion channels are not at all times possible, and when they are, they can be time-consuming. Using an artificial neural network (ANN) the aim is to predict the ionic currents during cardiac action potential (AP) in this work. A single-cell model was used to perform electrophysiological simulations to determine ionic currents based on ion channel conductance fluctuations. Then, the relevant ionic currents together with the related cardiac AP are calculated and fed into an ANN algorithm to anticipate the desired currents from the AP curve only. This thesis shows that when Bayesian regularization (BR) is utilized, the ionic current can be predicted with great accuracy from AP, as evidenced by the R (validation) scores of the whole data set, utilizing only AP forms.
Novel comp designs for improving the performance of cell- edge users in future wirless networks
The increase in cellular traffic is leading the advancement of cellular technologies to massively improve the quality of high-end communication experiences such as high throughput, high reliability, and universal access. To accomplish these demands in a cellular environment is a huge challenge, and it is more critical at the cell edges infected with interference. To address these requirements and challenges, we propose novel Enhanced Cooperative Multi-Point (eCoMP) and Hybrid Coordinated Multi-Point (H-CoMP) designs for cell-edge users to combat inter-cell interference (ICI), inter-user interference (IUI), and improve spectral efficiency. Superimposed supporting signals and channel-based precoders are designed to deal with ICI, remove channel effects, and provide reliable communication to cell-edge users. Mathematical models and simulations highlight the inventiveness and efficiency of the proposed paradigms compared to other competitive schemes. The results show the performance of the proposed solutions and the ability to reduce system complexity, leading to enhanced overall system performance.
Protecting IMSI from fake base stations exploitations and spoofers impersonations in 5G and beyond cellular networks
Fifth age (5G) portable organizations require confirmed and completely got transmissions between client gadgets and 5G base stations (gNodeB). In the ongoing 5G innovation standard, cell phones can't recognize a genuine and a phony base station (BS) by observing the guideline 5G security conventions. This weakness is a perilous security issue recognized in 5G cell networks around the world, where programmers and spoofers can undoubtedly imitate or hack the important data of portable clients. Albeit 5G security conventions are upgraded to forestall such sorts of assaults, counterfeit BSs compromise the security of remote correspondence organizations. To the best of the creators' information, there is right now no security technique accessible in the writing that can empower clients to distinguish who is sending the association demand reaction., in this work, we propose a clever hack-verification structure that can totally get IMSI bringing about avoidance against satirizing assaults.
Predicting the impact of cancer somatic mutations on protein-protein interactions
Cancer is a complex disease with one of the highest mortality rates. An existence challenge in cancer research is to identify the cancer driver mutation that causes cancer progression. Many biological processes are governed by protein-protein interactions. As such understanding which somatic mutations lead to a disruption in these interactions is critical for cancer research. A drawback common to most of the existing models is the assumption that a mutation in a protein causes the same effect on all of its interactions. Predicting how each interaction is affected by a mutation of interest is crucial for understanding disease development. In this thesis, we propose a model named Predator. Predator provides better prediction accuracy than the alternatives. The developed model is applied on the cancer data in the TCGA. Proposed approach is a novel study to analyze the effects of cancer mutations from TCGA on protein-protein interactions by classifying the impacts into disruptive/nondisruptive classes and having analyses specific to cancer. Such analyses may be beneficial for providing new insights regarding the mutations that are predicted to have a disruptive effect on cancer groups.
Comparative study of Pi and model predictive control of the dual active bridge DC-DC converter
The use of renewable sources requires suitable power electronics for integrating energy into the network. For this purpose, bidirectional converters become essential elements. Over the past years, a particular topology of bidirectional converter has attracted the attention of researchers due to its benefits: "The Dual Active Bridge converter". The operation and control of this topology must be thoroughly studied to increase its performance. In this thesis, a comparative evaluation of PI Control and Model Predictive Control of the DAB converter for use in a unidirectional power transfer application was conducted to choose the best control approach for the system operation. The methodology chosen for this analysis was the investigation of the system operation in three (3) different working scenarios: the steady-state, the dynamic response, and the input voltage change (disturbance). The predictive voltage control approach was developed as an S-Function for simulation in MATLAB.
Thermoelectric generator passive cooling system prototype design for use in photovoltaic panels
Today, as a result of the rising energy demand, the importance of alternative energy sources and the methods of increasing the efficiency of these is significant. One of these systems is thermoelectric generators (TEG), which can convert heat energy into electrical energy and operate in accordance with the Seebeck principle. These generators with P-N junction semiconductor structures, which can convert the temperature radiance between their two surfaces into electrical energy, are widely used in the heating/cooling systems, automotive and energy industries. It is seen in much research that waste heat energy has been converted into electrical energy by means of these generators and the energy needs of electronic devices with low power consumption have been met. In this study, the TEG design obtained by serial connection of Peltier coolers and the integration of this generator to the back sheet of the photovoltaic panel are investigated. The main purpose of the study is to determine whether the waste heat energy emitted by the photovoltaic panel can be converted into electrical energy by using it via a TEG. At the same time, it is the observation of the maximum efficiency of solar cells, whose ideal operating temperature is achieved by means of TEG passive cooling system. As a result of all these, it is observed that thermoelectric generators have energy production values of mW, even if methods such as heat sink and thermal paste are applied to effectively increase the temperature difference between the two surfaces of the generator. If the effect of the TEG on the passive cooling of the Photovoltaic panel is evaluated, it is obvious that there is no visible increase according to the tests made as a result of different days and hours. In the application phase, the methodology section is examined under two headings. In the first phase, a Peltier cooler design and thermal-electrical analysis are carried out using the COMSOL Multiphysics program. In the second stage, the prototype thermoelectric generator-PV panel system has been realized and power generation outputs are observed with the data taken at different days and hours. It is seen that the heatsink effectively increases the temperature difference between the two surfaces of the thermoelectric generator and provides up to 6% increase in generator energy production, especially on windy days. It is not possible to say that the same situation contributes to the passive cooling of the photovoltaic panel and that there is an increase in the amount of energy produced.
Investigating the use of impedance plethysmography for detecting decreased blood flow in diabetic patient limbs
The amount of blood flow into a body tissue is very critical and can be evaluated by monitoring the distention of the blood vessel (artery) feeding that tissue. Several techniques for artery distention monitoring have been reported in the literature. In this study, the impedance plethysmography (IPG) technique has been used as a non-invasive method with multiple electrodes to find the impedance across a finger, and therefore, monitor the existence of blood flow. Simulations have been performed on COMSOL finite element analysis software, and experiments have been carried out to obtain the optimum parameters for this technique. The simulations use 4 electrodes in different positions with a current of 10mA applied throughout a variety of frequencies up to 100 GHz. The resultant data clearly show the advantage of using multiple electrodes to enhance the total information received from the signals, while also indicating that 100Hz can be the optimum frequency for applying IPG. This study demonstrates that IPG with multiple electrodes can be applied to detect the existence of blood flow into body limbs, which is vitally crucial specifically for diabetic patients.
An investigation on energy harvesting from wrist for smart electronic devices
In this thesis energy harvested using the wrist movement of human arm is discussed. Human arm is constantly being used during our normal routine work, walking running or doing chores. These actions could be helpful in producing electricity. Previously research has been performed on the human body's ability to produce energy. Magnets have been utilized to design a device that harvests the energy using the wrist movement for electronic devices. The magnets were placed inside a 3-D printed tube and coils were wrapped the tube to convert the electromagnetic field into electricity. It can be worn to collect energy all day long. To determine the maximum performance throughout the arm movements, simulations were performed on software called COMSOL. The experiments were carried out by placing this device on the shaker and open circuit voltage was calculated with and without a resistor using an oscilloscope. The open circuit voltage generated at the least frequency of the shaker was 0.24 V and 0.064 V with resistance and without resistance, respectively. Different frequencies were applied to further measure the voltages. As batteries are constantly being needed to be replaced for the wearable electronic devices so, we developed the device which will continuously recharge them. This is a significant step towards future wearable electronics not requiring battery maintenance as it can charge the batteries as the wearer is normally doing their work in their routine.
Development of low-ion doped Cr:LiCAF lasers
Since the discovery of semiconductor lasers 50 years ago, these lasers have been found in numerous applications across various fields, including technology, scientific research, defense, healthcare, and more. These lasers required a great deal of aspects properties depending on the wavelength tunability and operation regimes. Based on this, for the femtosecond laser applications Ti:sapphire and Nd:YAG was one of the most prominent members of the solid-state laser family. Lately, as a low-cost alternative Cr:colquiirites lasers have been attracted. Cr:LiCAF laser, as a member of the colquiriite family, exhibits exceptional performance in terms of its physical properties, especially for power scaling applications. In this thesis, we have investigated diode pumped low ion doped Cr:LiCAF laser performance in terms of different laser operations. Power scaling potential of different lengths (10 mm, 12 mm, 15 mm, and 20 mm) Cr:LiCAF lasers tested with four multimode diode-pumped laser system. As a result, with a 10 mm 1.25% chromium doped gain medium 2.22 W output power was obtained. Continuous wave tuning range between 741-874 nm achieved. Afterward, in the gain switch experiments with a utilized pulse pumping setup 18% increase in the output power level was observed. We have also demonstrated the intracavity second harmonic generation. With different nonlinear crystals up to 441 mW, second harmonic power was generated. On the pulse generation parts, around 3 μs passively Q-switched Cr:LiCAF lasers were demonstrated. In passive mode-locking experiments, we have obtained a minimum of 65 fs pulses. Moreover, we have performed temperature estimation studies during lasing in Cr:LiCAF laser. Finally, off-surface birefringent filter plates two-color performance was tested on Cr:LiCAF laser numerically and experimentally. These studies show the potential of low-ion doped Cr:LiCAF laser for different applications.
Analysis of brain signals by EEG based emotion recognition methods
This study examines an experiment conducted at Akdeniz University and Antalya Bilim University, aiming to evaluate students' responses to different drawing methods. The experiment was carried out using neurophysiological tools, such as EEG (electroencephalography). EEG, as a technique for measuring electrical activity in the brain, was employed to assess the emotional states experienced by participants during drawing. These neurophysiological data helped in understanding how students reacted while using different drawing methods and how these methods influenced their emotional states. The results particularly highlight the positive impact of the drawing experiment using a 3D program. This suggests that technological tools and new software can enhance students' motivation. The findings of this study provide valuable insights for the selection of drawing tools and the design of educational programs for students in the field of interior architecture. Additionally, it sheds light on the potential use of neurophysiological tools in evaluating emotional states.
A comparative analysis of deep learning architectures for breast cancer detection in ultrasound imaging
This study investigates the area of breast cancer classification using deep learning models, with a specific focus on examining the efficacy and possible improvements of Convolutional Neural Networks (CNN), U-Net, and ResNet50. The analysis of these models is conducted on a breast ultrasound dataset to evaluate their respective merits, limitations, and technical complexities. The CNN model demonstrates modest accuracy and exhibits a relatively basic architecture, contrasted with the U-Net model which achieves outstanding accuracy, particularly in the identification of normal cases. Conversely, ResNet50 achieves an equilibrium between sensitivity and specificity, demonstrating effective adaptability to a broad range of features within datasets. The training dynamics observed in each model provide valuable insight into their respective learning capabilities. Specifically, the CNN model demonstrates consistent convergence, the U-Net model exhibits rapid learning, and the ResNet50 model displays balanced convergence. These findings offer important information regarding the performance and behavior of these different models in the context of machine learning and image processing tasks. In conducting a critical analysis, their performances are compared with particular emphasis placed on the significance of precision and recall in clinical applications. The technical analysis provides an in-depth examination of the intricacies within architectural structures, data complexities, and training dynamics, ultimately providing a comprehensive understanding of the behavior exhibited by each model. The study proposes future research to improve the existing models through various means, including dataset augmentation, architecture refinement, transfer learning, clinical validation, explainability, and ensemble approaches. The imperative of interdisciplinarity is underscored in facilitating the transition of these models, initially developed as research prototypes, into practical and clinically significant tools for medical imaging and the diagnosis of breast cancer. This necessitates the collaborative effort of data scientists, clinicians, and imaging experts. In summary, this study makes a valuable contribution to the comprehension of deep learning models in the classification of breast cancer, while also offering a framework for prospective advancements. This work facilitates the convergence of technical innovation and clinical applicability in the field of medical imaging.
Optimizing photovoltaic solar systems for sustainable rural electrification
The research focuses on Tehsil Saleh Pat, Goth Gagrawara, District Sukkur, Sindh, Pakistan, to assess the benefits of photovoltaic solar power electricity in rural regions. Another goal is to investigate the public's opinions, attitudes, and sense of ownership toward a specific project conducted in the village, as well as to discover the underlying causes of these sentiments. Furthermore, the study aims to analyze and grasp rural populations' energy usage habits. Conducting interviews with both beneficiaries and non-beneficiaries in the village of Gagrawara, arranging two focus group talks with females, and monitoring behavioral trends within the village were all part of the data collection for this study. In addition, the research intends to evaluate the performance and operation of an existing grid-connected PV power plant in Gagrawara, Sindh. This PV system has a power rating of 8.6 kWp, and extensive meteorological and performance data measurements have been collected for all operational years. The appraisal of this PV facility is extremely important for the PV business in Sindh. Furthermore, a fundamental goal of this thesis is to improve our understanding of PV system performance and operation in Sindh, therefore offering useful insights to the PV industry's development. The reason for selecting Gagrawara village as the location for this study is the ongoing implementation of the Photovoltaic solar electricity project by Pentera Energy. This study's project is intended to reach at least 60–70 households. The project has already begun to, directly and indirectly, benefit the residents of the village. According to the study, remote rural areas without access to the main grid have found Photovoltaic solar power electricity is a dependable source of energy. It is the ideal method for promoting development initiatives and generating opportunities for people who live in rural areas. The findings of this research indicate that the local community holds a positive view toward the installation of photovoltaic solar power systems in their village. The driving force behind this conduct is predominantly linked to the prerequisites that must be satisfied before the project can be executed, along with the benefits that the community has already enjoyed. Most residents of the village now have more time for socializing and shopping, and they can exchange ideas more freely in the evening and at night. There is no longer any question about PV solar electricity's efficiency in rural communities. In light of the challenges faced by rural areas in contributing to the process of development, it represents the optimal choice for providing electricity to these regions. Because it offers a substitute energy source that can be advantageous to many people in underserved areas because of its accessibility, cleanliness and affordability, and safety. Energy stakeholders can no longer ignore the significance of PV solar energy. As a result, it is important to support local businesses and other shareholders in their efforts to further engage with rural poor communities.
Deep learning technique for early detection of lung cancer
Cancer is a deadly illness brought on by a confluence of many metabolic anomalies and genetic disorders. One of the main causes of death and disability among people nowadays is lung cancer. When choosing the appropriate course of action, the histopathological identification of such malignancies is typically the most crucial factor. Early diagnosis of a disease on either front significantly reduces death rates. Researchers can investigate a large number of patients in a much shorter amount of time and at a cheaper cost by using machine learning and deep learning techniques to speed up such cancer diagnosis. To effectively detect lung cancer, we developed a hybrid ensemble feature extraction model in this study. It combines high-performance filtering for cancer imaging collections with deep feature extraction and ensemble learning. Histopathological (LC25000) lung and colon datasets are used to assess the model. The results of the study show that our hybrid model has a 99.63% accuracy rate in detecting lung cancer. The results of the investigation demonstrate that our suggested approach performs noticeably better than current approaches. These models may therefore be used in clinics to assist physicians in diagnosing cancer. KEYWORDS: Deep learning, Feature extraction, Hand-Crafted Descriptors, lung cancer, and Transfer learning.
Brain tumor segmentation using deep learning approach
The research under consideration studies the Brain tumor segmentation using deep learning approach. The goal of the research is to discover how the models' layer parameters and the hyper-parameters changes are the parameters that affect the models' accuracy and performance. This paper was motivated by the pressing need for medical imaging to be done with utmost accuracy, hence, the main goal of this research is to find the U-Net architecture design that will enable the brain tumor segmentation to be done with the highest possible accuracy. The systematic procedure was carried out, which consisted of the collection of a large number of brain tumor cases in an unbiased way. The models were firstly trained and then evaluated, on the other hand, the attention was mostly on the key performance indicators. The study's findings indicate the minor differences in model performance and accuracy which are most probably due to the various architectural sets. It is worth to mention that some of the configurations were more stable and accurate in segmentation, which shows that the architectural design is the base in the medical image processing tasks. The findings reveal how the architectural modifications that are made for the patients serve the purpose of the segmentation of the brain tumors. The research findings are giving the medical image segmentation practitioners and researchers an important direction and a new area for growth in the field.
A comparative study of YOLOv8 and faster R-CNN in fruit leaf disease detection for precision
This study investigates the use of deep learning models, specifically YOLOv8 and Faster R-CNN, for real-time object identification and autonomous plant disease diagnosis. The research uses a dataset consisting of 38 examples from 11 distinct classes, which include various plant species such as apple, grape, and potato. Within these categories, there are specific classes representing healthy conditions and various diseases.A statistical analysis of the subset of the dataset reveals different distributions of images across the classes, highlighting the prevalence and significance of certain diseases within each plant category. The effeciency of YOLOv8 and Faster R-CNN is assessed using performance metrics like Intersection over Union (IoU), saliency score, and inference time. Although specific numerical values are not provided, the data indicate that YOLOv8 performs well in terms of IoU and achieves a higher saliency score compared to Faster R-CNN. Conversely, Faster R-CNN shows superior IoU performance but with a lower saliency score. Additionally, YOLOv8 demonstrates faster inference times, while Faster R-CNN has significantly longer inference times.By comparing these metrics, the study provides valuable insights into the strengths of each model, guiding the selection and optimization of deep learning architectures for plant disease diagnosis. The research also emphasizes the trade-offs between speed and accuracy inherent in object identification models, underscoring the importance of considering application-specific requirements. Overall, this study advances agricultural technology by exploring the potential of deep learning models in combating harmful plant diseases. It lays the groundwork for future advancements in autonomous plant disease diagnosis and contributes to global food security by enhancing diagnostic techniques.
Differential privacy for machine learning
This study examines an experiment conducted at Antalya Bilim University, Using machine learning algorithms to predict disease risk is a promising way to improve patient outcomes and early diagnosis in healthcare and predicting disease risk is a promising way to improve patient outcomes and early diagnosis in healthcare. Our results provide a high-level comparative visualization of six machine learning models using confusion matrices and their respective accuracy scores. The models evaluated are Logistic Regression, Random Forest, K-Nearest Neighbors (KNN), Naive Bayes, Decision Tree, and Support Vector Machine (SVM). The confusion matrices display the performance of each model by showing the breakdown of actual versus predicted classifications for four distinct classes. Each model is assessed in terms of its ability to correctly classify instances into these classes, with the diagonal of each confusion matrix representing correct predictions (true positives and true negatives). The intensity of the colors in the matrices indicates the concentration of predictions in each category. Above each matrix, the overall accuracy of the model is displayed, which quantifies the percentage of correct predictions across the entire dataset. This visual comparison highlights not only the accuracy of each algorithm but also gives deeper insights into the types of errors made (such as misclassifications between similar classes). The results suggest that while all models achieve relatively high accuracy, certain models like SVM and KNN slightly outperform others in terms of overall precision, minimizing misclassifications in specific class categories.
Enhancing machine learning algorithms in healthcare with electronic stethoscope
In this study, our aim is to classify respiratory sounds and diseases via audio and text data recorded by an electronic stethoscope using convolutional neural networks (CNNs), support vector machines (SVMs), k-nearest neigbor (k-NN) and Gaussian Bayes (GB) algorithms on a dataset that contains 17,930 lung sounds that were recorded from 1630 subjects. For classifying respiratory sounds, we employed; SVM, k-NN and GB with mel frequency cepstral coefficient (MFCC) features and CNN with 28x28 and 600x600 spectrogram images. We prepared 4 datasets to classify respiratory audio into: (1) healthy versus pathological; (2) rale, rhonchus, and normal sound; (3) singular respiratory sound type; and (4) audio type with all sound types classification. Accuracy results in percent were; (1) CNN 86 and 95, SVM 86, k-NN 85, GB 58, (2) CNN 80 and 93, SVM 80, k-NN 79, GB 42, (3) CNN 76 and 85, SVM 75, k-NN 76, GB 22 and (4) CNN 62 and 77, SVM 62, k-NN 61, GB 15 respectively. For classifying respiratory diseases, SVM, k-NN and GB algorithms were run on 6 datasets to classify patients into; (1) ill or healthy with text data, (2) ill or healthy with audio MFCC features, (3) ill or healthy with the text data and audio MFCC features, (4) 12 diseases with text data, (5) for 12 disease with audio MFCC features, (6) for 12 disease with the text data and audio MFCC features. Accuracy results in percent for SVM were 75, 88, 64, 73, 63, 70; for k-NN 95, 92, 92, 67, 64, 66; for GB 98, 91, 97, 58, 48, 58 respectively. To compare the electronic and traditional stethoscope, 3 chest physicians assessed 100 audio clips. We observed; good level consistency between physicians 2 and 3, average level consistency between physicians 1, 3 and 1, 2 via kappa statistic method.
Ontology based data access with relational databases
As the interconnected devices on the Internet increases around the world, it becomes increasingly important to gather and share meaningful data, also make automatic inferences. As a result, uniform access to stored data, especially in Relational Database Management Systems (RDBMS), is a problem for next-generation web technologies (i.e. Semantic Web) and Internet of Things (IoT). An ontology defines a high-level global schema of data sources and provides an intelligent vocabulary for user queries. Ontology Based Data Access (OBDA) has become an important paradigm for accessing stored data in heterogeneous sources and regarded as a key ingredient of the next-generation of information systems. In this study, stored data in a relational database system were accessed over OBDA model through SPARQL queries according to developed ontology. After developing an ontology that describes the items and relations in the database, mapping process was applied between ontology instances and database entities in a semi-automatic fashion. Equivalent SPARQL queries are automatically generated for the SQL queries sent via the client software and the relational database was accessed. The success of the OBDA procedure was verified by cross-comparison of the data between the relational database and data returned from our model. The response time for each SQL and SPARQL query was also compared. This study may be considered as an authentic solution example to one of the obstacles accessing stored data in relational database system over an ontology for next- generation of internet technologies. All of the tools and the software have been used in this research are also entirely free and based on open source technology.
Performance analysis of the ellipsoidal support vector clustering algorithm on various synthetic and biomedical data sets
Along with the development of communication and technology, the amount of digital data in the world is increasing. Therefore, it has become very important to obtain meaningful information by using these data especially in recent years. In particular, clustering methods are one of the most important algorithms used to extract hidden patterns within the data. In most clustering methods, the Euclidean distance is used as a similarity metric between data samples. At the Euclidean distance, the variance of the data is considered equal. However, most real world data may contain data in different variances. For this, kernel-based Ellipsoidal Support Vector Clustering (ESVC) algorithm was developed using Mahalanobis distance. The ESVC method can automatically generate the appropriate cluster boundaries according to the data samples without having to specify the number of clusters by means of the variance parameter of the kernel function. This study is divided into two parts. In the first part, the ESVC method was first time applied to the real biomedical data sets such as Hepatitis and Parkinsons, and then was compared with clustering algorithms that use Euclidean distance such as k-means, fuzzy c-means and hierarchical. From the simulation results, it was observed that the ESVC algorithm performed quite well. In the second part, Cauchy, Laplacian, hyper tangent kernels were proposed as an alternative to Gaussian kernel for ESVC and then performance analyzes were performed. According to the results of the analysis, it was shown that Cauchy and hyper tangent kernel functions could be used as an alternative to Gaussian kernel.
Design of induction generator for wind turbine
The electromagnetic induction method is usually used for this purpose. Electromagnetic induction is carried out by moving a wire file around a fixed rod with an electric field, either by a magnet or an electric magnet. The induction generator depends on the principle of electromagnetic induction. This phenomenon was discovered by Michael Faraday, and it is reported that the current is produced in the conductor when it is moved through the magnetic field. The resulting output voltage is called the electromotive force (emf). This process can also be reversed. In other words if the electric charge is moving, it will generate a magnetic field. The Faraday-Maxwell induction code determines the relationship in which the value of emf versus the number of cycles (for a file) indicates the rate of change in the magnetic flow. The reason for the minus sign is that the direction of the wattage direction is opposed to the change in flow that produces the voltage. Electromagnetic induction has many applications. For example, in credit cards (magnetic strips on the back of the card) or in order to generate voltage in the motor. Another one is very important because the generator is an "inverted" engine. The objective of this study is to design a small wind turbine which is optimized for the constraints that have residential use. The design process includes the selection of the wind turbine type and the determination of the blade airfoil pitch angle and chord length distribution along the radius. The pitch angle and chord length distributions are optimized based on conservation of angular momentum and theory of aerodynamic forces on an airfoil. Blade Element Momentum (BEM) theory is first derived and then used to conduct a parametric study that will determine if the optimized values of blade pitch and chord length create the most efficient blade geometry. Finally, two different airfoils are analyzed to determine which one creates the most efficient wind turbine blade. The study includes a discussion of the most important parameters in wind turbine blade designed to maximize the efficiency.
Ultrasonografide karaciğer yağlanmasının nicel olarak değerlendirilmesi
Karaciğer yağlanması günümüzde sıkça karşılaşılan hastalıklarından bir tanesidir. Bu hastalığın teşhis ve takibinde ultrason, bilgisayarlı tomografi, manyetik rezonans görüntüleme gibi görüntüleme tekniklerinin yanısıra biyopsi ve kan testleri de kullanılmaktadır. Çeşitli alanlarda teknoloji gelişmeye devam ederken, hastalıkların teşhisinde, hâlâ maliyeti düşük, kolay erişilebilir, hızlı ve invaziv olmayan yöntemler öncelikli olarak tercih edilmektedir. Karaciğer yağlanmasının tespit ve takibinde ise en çok kullanılan yöntem ultrason görüntüleme yöntemidir. Ultrason kolay ulaşılabilen, düşük maliyetli ve insan sağlığı açısından bilinen zararı olmayan bir yöntemdir. Ultrason görüntüleme yöntemi karaciğer yağlanmasının tespitinde oldukça sık kullanılmasına rağmen ultrason görüntülerini kullanarak hastalığı teşhis etmek için her zaman bir uzman görüşüne ihtiyaç duyulmaktadır. Uzman, karaciğer yağlanmasını, ultrason görüntülerindeki bulguları değerlendirerek derecelendirmektedir. Ancak, ultrason görüntüleri üzerinde uzman tarafından yapılan bu karaciğer yağlanmasına ilişkin değerlendirme farklı zamanlarda farklı yorumlara sebep olabilmektedir. Bir diğer ifadeyle, karaciğer yağlanma seviyesinin bir öncekine göre farklı seviyede olduğu yönünde (subjektif) yorum yapabilmektedir. Bu tezin amacı, ultrason görüntüsünden karaciğer yağlanmasının objektif olarak tespit edilmesinde uzman görüşünü destekleyen ve uzmana yardımcı olabilecek matematiksel bir yöntem önermektir. Böylece önerilen yöntem ile elde edilen nicel sonuç sayesinde hem hastaya daha objektif bilgi verilebilecek, hem de ultrason görüntülerinin farklı zamanlarda değerlendirilmesinde uzmana kolaylık sağlanabilecektir.
Low cost and flexible terminal node controller design
With the spread of chaos and instability in conflict areas, the need for a tracking system has increased. In this thesis, a design and implementation of a low-cost communication network is introduced. The designed system uses UHF/VHF for vehicle localization using custom APRS hardware built with off-the-shelf electronic components and registration of acquired location using local database service. The proposed system also takes into account the financial aspect and accessibility of equipment as a prerequisite for the success of the network.
An intelligent use of stem and morphology analysis in arabic information retrieval
In the past several years, the Arabic information retrieval has garnered significant attention due to increasing the Arabic text on the web. A considerable number of researchers share similar opinions on the benefits of morphology and stemming in Arabic information retrieval systems, especially for internet search engines; a problem exacerbated by the enormous amounts of datasets on the internet. The Arabic language is ranked as the seventh top language on the web. It is the highest growth of the ten top online languages. Therefore, the number of Arabic documents increases rapidly. Also, the Arabic language has a serious challenge due to the complexity of its alphabet morphological. In NLP tasks it becomes hard to select an effective index term of information retrieval systems. Thus, indexing terms is a complex and difficult process, especially when it concerns the indexing of Arabic documents. Year after year, many methods are being published to overcome the Arabic stem problem for successful retrieval of documents. Therefore, this research present a novel method to extracting an Arabic stem called Arabic Morphology Information Retrieval (AMIR. The main goal and advantage of our method are to generate/extract stem by applying a set of rules and matches the relationship between some Arabic letters to find the root/stem of the respective words to use as indexing terms for the text searching in Arabic retrieval systems. Furthermore, we highlight the use of these rules and their benefits for different Arabic information retrieval systems. Consequently. AMIR can be considered to operate around minimum morphological complexity. Finally, AMIR has been tested using the EveTAR (2016) dataset on Arabic tweets and the obtained results show that the AMIR results outperform the state-of-the-art results. Therefore, our approach has been able to improve the performance of Arabic stem and increases retrieval as well as being active against any type of stem and we believe that it's difficult to develop a new Arabic system retrieval method without uses a good morphology analysis support it.
Sıft anahtar noktalarını kullanarak ham video bileşenlerinde veri gizleme ve gizli verileri tespit etme
Bu çalışmada, veri gizleme yöntemlerinin önemli bir alt disiplini olan steganografi ve steganaliz üzerinde durulmuş ve farklı yöntemler kullanılarak elde edilen sonuçlara göre çıkarımlar yapılmıştır. Taşıyıcı ortam olarak hem zamana hem de uzaya yayılmış gerçek zamanlı olarak alınan ham video bileşenleri kullanılmıştır. Öznitelik çıkarım algoritmalarından SIFT kullanılarak veri gizlenen gerçek zamanlı ham video bileşenleri için, hem istatistiksel/matematiksel modellemede piksel komşuluk matrisine dayalı hem de derin öğrenmede CNN detektörleri kullanılmasına dayalı steganaliz çalışmaları yapılmıştır. Taşıyıcı ortama veri gizleme işlemi yapılırken, gerçek zamanlı ham video bileşenlerinin her bir çerçevesi için ayrı ayrı elde edilen en kaliteli 5 (beş) SIFT anahtar noktasından mükerrer olmayan noktalar kullanılmıştır. Hem veri gizlenmeden önceki örtü dosyalarından hem de veri gizleme işlemi yapıldıktan sonraki örtülü dosyalar içerisinden ayrı ayrı elde edilen SIFT anahtar noktalarının uzay düzleminde birebir aynı olduğu ve aralarında çok ciddi yapısal farkların olmadığı tespit edilmiştir. Bu SIFT anahtar noktalarına, bit uzayında LSB yöntemi kullanılarak veri gizleme işlemi yapılmıştır. SIFT yöntemi ile elde edilen anahtar noktalara yapılan veri gizleme işlemi sonucunda örtü medyaya saklanan gizli mesaj ile örtülü medyadan elde edilen gizli mesajın birebir aynı olduğu görülmüştür. Bu çalışma, gizli mesajı içerik olarak bulmayı ve gizli mesajın içerik olarak bulunup yok edilmesi ve/veya değiştirilerek bozulmasını amaçlamamıştır. Sadece istatistiksel/matematiksel ve derin öğrenme tabanlı olarak gizli mesaj içeriğinin var olup olmadığı bilgisini tespit etmeyi hedeflemiş ve elde edilene sonuçlar değerlendirilmiştir.
Retinoblastom hastalığında yeni nesil dizileme veri analizi ile bir ardışık düzenin geliştirilmesi
Yeni nesil dizileme (YND), genomik araştırmalarda devrim yaratan büyük ölçüde paralel deoksiribonükleik asit (DNA) dizilemeye dayalı bir teknolojidir. YND verisi oluşturmanın maliyeti, bu teknolojinin ortaya çıktığı zamandaki ile karşılaştırıldığında azalmış olsa da, çalışılan veriye göre yine de bir problem oluşturabilir. Bu maliyet sorununun üstesinden gelebilmek için havuz dizileme ve düşük kapsamalı dizileme verileri gibi yeni stratejiler geliştirilmiştir. Düşen maliyete rağmen, bu stratejilerin YND çalışmalarında etkili olup olmadıklarını değerlendirebilmek önemlidir. Bu tez kapsamında, tümör verilerinden elde edilen havuz dizileme ve düşük kapsama ile dizilenmiş retinoblastom verisi için bir ardışık düzen geliştirilmiştir. Retinoblastom, çocukluk çağında RB1 mutasyonu veya MYCN amplifikasyonu ile başlayan ve göz (ler) in görme kaybına ve hatta bazen ölüme yol açabilen bir göz kanseridir. Geliştirilen ardışık düzenin etkinliğini değerlendirebilmek için, hem bu özellikteki hastalık verisi hem de mümkün olabildiğince benzer özellikteki diğer veriler üzerinde standart bir ardışık düzenle birlikte çalışılarak sonuçlar karşılaştırılmıştır. Geliştirilen ardışık düzenin, daha fazla sayıda varyant çağırabildiği ve daha yüksek duyarlılık ve F-skor değerleri elde ettiği gözlemlenmiştir. Ek olarak, retinoblastom verisinde varyantlar, hastalık ile ilişkili genlerde çağrılan varyantlar ve varyant türleri ile ilgili sonuçlar sunulmuştur. Geliştirilen ardışık düzenin etkinliğini daha net değerlendirebilmek adına, daha yüksek mutasyon oranlarına ve daha büyük havuzlara sahip kanser verilerinin kullanılması önerilmektedir. Öte yandan, YND veri analizinde hizalama adımı hem zamana gereksinim duyan hem de bu adımın temel karakteristiği sebebiyle GPU'ya uyumlu olduğundan, CPU'da çalışan hizalama algoritmalarının bazılarının GPU'da çalışabilen versiyonları geliştirilmiştir. Geliştirilen ardışık düzen kapsamında kullanılan hizalama algoritması olan BWA ve GPU için CUDA ortamında geliştirilen versiyonu olan BarraCUDA farklı veri setleri üzerinde incelenerek, performansları değerlendirilmiştir. Buna göre, BarraCUDA'nın tek GPU üzerinde bile ciddi anlamda çalışma süresini azalttığı ve çalışılan tüm verilerde belirtildiği gibi BWA ile benzer hizalama oranına sahip olduğu gözlemlenmiştir. GPU'nun katkısının tam olarak anlaşılabilmesi için, BarraCUDA'nın okuma sayısı fazla olan veriler üzerindeki yaklaşımı ve birden fazla GPU kullanımının etkisi incelenmelidir.
Modeling and intelligent control of a standalone PV-wind-diesel hybrid system
The aim of this thesis is to study the modelling and intelligent control of a stand-alone hybrid energy system based on solar-wind-diesel with battery. The renewable sources are major components of a standalone hybrid system as a combining photovoltaic with the wind turbine. Each component of these systems has been modeled and implemented in MATLAB/Simulink software. The proposed system is able to operate in standalone mode, without the need for connection to the main grid. The fluctuations in voltage and frequency of the system can be encountered depending on wind speed, solar irradiation and the temperature. Thus it can be observed the decrease in power quality in such systems which operate in standalone. This condition can even go worse when varying loads are considered. In order to improve power quality and reduce unwanted fluctuations of system quantities under different operating conditions an intelligent control is designed in this work. The capabilities of this designed intelligent control system on the control of voltage and frequency of the system are tested under different operating conditions and compared to PID controllers. The superiority of the designed intelligent control system over PID controllers is shown by the simulation studies.
Smart location-based mobile shopping Android application
In this thesis, a standard feature of today's smart phones "the location information" is used for determining the user location. SAGO finds nearest electronic super-stores by using the location information. Currently, SAGO shows only the electronic malls however, it will be added other types of shopping stores such as clothing stores, perfumery and so on. Therefore, this mobile shopping Android application's coverage is not limited. The application is easily extendable according to the users' requirements. The Android application is developed in Android 2.2 version, Application Programming Interface (API) level 8. Although the mobile application is developed for Android devices, it will be easily adopted and developed for iOS devices. The system architecture is designed by using Service-Oriented Modeling Framework (SOMF) which is based on the Unified Modeling Language (UML) and also, object-oriented programming language is used in the development process. Designing with UML is simplifying the complexity of the system and helps us to understand the architecture of this thesis project. In the development process, the users' input data which has been brought before searching products in nearest electronic super-stores and also the names of local shopping stores are used in a "Smart Filtering algorithm". Moreover, simple filtering and cleaning technics such as Agglomerative Clustering Algorithm, Greedy Search Algorithm, and a Levenshtein distance are used in Search And Go (SAGO) mobile shopping application. By using this Android mobile application, the users are able to search the products in closer areas. The mobile users do not need to know where the locations of the shopping malls are. Also, the users are able to compare the prices that they are getting from each local electronic super-store and subsequently decide where to buy. The product is not purchased thorough the SAGO mobile shopping application. If the mobile users want to buy products, they will go to the related store directly. According to these, the users save both their money and time by using this Android mobile application.
Indoor Location Estimation by Using Maximum Likelihood Estimation Based Algorithm on Small Cell Networks
This thesis presents a new framework for indoor localization using Third Generation(3G) Universal Mobile Telephone System (UMTS) Femtocell. The fingerprinting technique is applied to collect the Received Signal Strength Indication(RSSI) values through an Android User Equipment(UE) and data is processed in real time using MQTT server. To achieve better Radio Frequency (RF) planning and optimization for the placement of Femto Access Point(FAP), statistical analysis is performed by normalizing and calculating the Mean Squared Error(MSE) of the acquired data. To maximize the success rate in finding the location of the person, Maximum likelihood Estimation(MLE) based algorithm is used for tracking. Simulation was carried out both for 1 million samples and real life test using 100 samples. To make it more accurate and efficient, MLE based algorithm is developed and simulated in MATLAB. Both simulation approaches resulted in good success rates. A case study is also part of this thesis, case study discuss about the real life project and deployment setup for indoor positioning estimation and localization techniques using multiple FAP's.
Evaluation of e-government project in Iraq: the general directorate of traffic (Case study)
The main objective of the e-government project is to provide online services to citizens by using the information technology and particularly the Internet. Thus, the use of information and communication technology (ICT) is the key element of any e-government project. Consequently, great efforts are presented by governments to develop their infrastructure, staffing skills, governmental websites and citizens' awareness to achieve this objective. In Iraq, the government took its first steps to develop the e-government at the end of 2003. These steps represented by putting the implementation strategy of e-government with the help of the United States. Most ministries and institutions had create their own e-government project through their websites. Although it provided few services for citizens, the initiatives were likely good as a young project. Unfortunately, there were many challenges that handicapped the development of the e-government in Iraq involving political reasons, poor infrastructure, digital divide and lack of funding lack. The last reason played an important role in the middle of 2014 and the stakeholder could not maintain the portal of the Iraqi e-government. However, some governmental institutions still create or develop their own e-government projects. This thesis makes an evaluation of the e-government project of the General Directorate of Traffic (GDT) as case study by using a risk assessment technique called Design-Reality Gap (DRG) model to avoid failure in such project we made. This evaluation introduces the gap exist between the design stage and the reality situation in the GDT. Besides, it takes actions to reduce the gap to increase the success rate. In addition, a proposed assessment technique used to complete the evaluation process. This thesis also presents the e-government of Iraq in general as well as the e-government of two other countries. During the evaluation process, the thesis emphasizes on the technical aspects of the e-government project.
Estimation of commercial building energy consumption with machine learning
The estimation of energy consumption in commercial buildings holds immense significance in the pursuit of sustainable energy management and the efficient allocation of resources. As energy demands continue to rise, optimizing energy consumption becomes crucial for reducing carbon footprints and enhancing cost-effectiveness. In response to these challenges, this research endeavors to leverage the power of machine learning (ML) techniques to accurately predict energy consumption in commercial buildings. By employing ML algorithms, this study seeks to improve energy efficiency, lower operational costs, and facilitate informed decision-making in building energy management. To achieve these objectives, the research employs eight ML algorithms, namely Support Vector Machines (SVM), Random Forest (RF), Extra Trees, Linear Regression, Lasso, Gradient Boosting Regressor (GBR), Multilayer Perceptron (MLP), and a novel stacking model. Each of these algorithms is well-known for its predictive capabilities and ability to handle various types of data. The research methodology encompasses the development and assessment of predictive models using a robust and extensive dataset, carefully collected from commercial buildings. Before applying the ML algorithms, the dataset undergoes rigorous preprocessing to ensure data quality and enhance model performance. Normalization, data cleaning, and feature selection techniques, such as ANOVA and Relief F, are employed to eliminate noise and irrelevant information, making the dataset suitable for training and evaluation. The research splits the data into training and testing sets in a balanced ratio of 70:30, ensuring that the models are trained on a substantial portion of the data while being evaluated on unseen samples. ii This balanced data splitting allows for unbiased evaluation and comparison of the performance of the ML algorithms. Remarkably, the results reveal that the stacking model, a novel approach that synergizes multiple models' strengths, outperforms all other ML algorithms and state-of-the-art approaches. The stacking model showcases an impressive Mean Absolute Error (MAE) of 0.01286, Mean Squared Error (MSE) of 0.00054, and Root Mean Squared Error (RMSE) of 0.02328. These exceptional performance metrics demonstrate the stacking model's potential as a highly effective tool for accurately estimating energy consumption in commercial buildings. The research findings are further validated by comparing the results with related works in the field. The stacking model consistently outperforms existing approaches, reaffirming its superiority in accurately predicting energy consumption. This model's success can be attributed to its ability to combine diverse models, leveraging their complementary strengths to capture complex relationships within the energy consumption data. The adoption of the stacking model as the best-performing ML algorithm in energy consumption estimation carries significant implications for sustainable energy practices in the built environment. Accurate energy consumption prediction empowers building managers to make informed decisions on energy allocation, implement energy-saving measures, and optimize energy usage. By enabling data-driven energy management, the stacking model supports the advancement of sustainable energy practices, contributing to the reduction of carbon emissions and the promotion of environmentally friendly building operations. In conclusion, this research showcases the immense potential of machine learning techniques in accurately estimating energy consumption in commercial buildings. The stacking model's exceptional performance, when compared with other ML algorithms and state-of-the-art approaches, highlights its role as a transformative tool in achieving energy efficiency and sustainable energy management. As the world continues to prioritize sustainability, the implementation of the stacking model can significantly contribute to building a greener and more energy-efficient future.
User behavior analysis on e-commerce using NLP techniques
This study presents an in-depth investigation into the potential for sentiment analysis (SA) and machine learning (ML) in facilitating the sales prediction and customer retention processes for both small and large-scale businesses. Online platforms including blogs, social networks, and review portals have transformed the marketing landscape, enabling consumers to voice their opinions on a vast array of topics, from product reviews to popular culture. These digital platforms not only foster customer engagement, but also provide businesses with an invaluable data source for predictive analysis, essential in strategic sales forecasting and customer relationship management. In this study, we compiled a comprehensive dataset of product review tweets, serving as a rich representation of consumer sentiment. To ensure the integrity and relevance of the data, we engaged in rigorous preprocessing methodologies, mitigating potential noise and inconsistencies. Following the cleaning phase, we utilized the Valence Aware Dictionary and sEntiment Reasoner (VADER), a lexicon and rule-based sentiment analysis tool, in conjunction with several machine learning algorithms. The objective was to ascertain the most effective means of classifying the sentiment polarity of product reviews, subsequently aiding in sales prediction. Our findings reveal that while VADER offers notable benefits in sentiment analysis, ML techniques present superior accuracy in classifying the polarity of product reviews. More specifically, logistic regression (LR) was found to be the top-performing algorithm in this context. Across a multitude of evaluation metrics, including accuracy, precision, recall, F1-score, Matthews correlation coefficient (MCC), and area under the curve (AUC), LR consistently outperformed its ii counterparts, thus solidifying its position as the most apt choice for sentiment classification in product reviews. Notably, other algorithms such as XGBoost and Stochastic Gradient Descent (SGD) also demonstrated competitive performance. They can serve as plausible alternatives in situations where model interpretability is not the prime concern and a higher degree of model complexity is permissible. These findings contribute to an emerging body of knowledge, illuminating the potential of SA and ML in providing businesses with robust tools for understanding customer sentiment, predicting sales, and consequently enhancing customer retention strategies. The implications of this study extend beyond academia, promising substantial real-world benefits for various stakeholders in the business sphere.
Predicting the value of football player with the impact of covid-19 on the market value of the player
The unprecedented onset of the Covid-19 pandemic has pervaded countless industries worldwide, with football being a major one to bear its brunt. This global health crisis ushered in a new era where normalcy was challenged and socio-cultural constructs were altered, leading to profound implications on football, both as a sport and an industry. Football, a nexus of economic and cultural intersections, witnessed several changes, not least among which was the effect on player valuations in the market. The pandemic brought life to a standstill, ensnaring societies in its clutches and forcing them into isolation. This sudden cessation had a ripple effect on football players, amplifying feelings of anxiety, tension, and uncertainty, thus altering their psychosocial dynamics. Consequentially, this shift in psychological state, combined with the disruption of regular football activities, contributed to fluctuations in the market prices of these athletes. Traditional approaches and methodologies, once relied upon for gauging player prices, particularly in the five major European football leagues (English, Spanish, Italian, German, and French), began showing signs of obsolescence in the face of these pandemic-induced challenges. So, this study presents the performance assessment of five machine learning algorithms—Linear Regression, Lasso Regression, Random Forest, Gradient Boosting, and K-Nearest Neighbors—on a regression problem. The evaluation metrics utilized are Mean Absolute Error (MAE), Root Mean Absolute Error (RMAE), and R-squared (R2). Linear Regression demonstrates a high R2 of 0.97 but falls short in MAE and RMAE. Lasso Regression surpasses with an R2 of 0.99 but exhibits the highest prediction errors, hinting at possible overfitting. Random Forest provides a balanced outcome with a R2 of 0.91 and moderate errors. Gradient Boosting stands out in terms of prediction accuracy, having the smallest MAE and RMAE, despite an R2 of 0.90. Meanwhile, K-Nearest Neighbors parallels Linear Regression's performance. The study suggests that Gradient Boosting might be the most accurate for predictions based on MAE and RMAE. However, the ideal model selection is contingent upon the specific objectives of the regression problem, which could prioritize explaining variance or minimizing prediction errors.