Bilgisayar Mühendisliği Bölümü / Department of Computer Engineering
Permanent URI for this collectionhttps://hdl.handle.net/11413/6817
Browse
Browsing Bilgisayar Mühendisliği Bölümü / Department of Computer Engineering by Rights "info:eu-repo/semantics/restrictedAccess"
Now showing 1 - 20 of 22
- Results Per Page
- Sort Options
Publication Adaptation of Gamification as a Man-Machine Interface to Franchise Management System(Institute of Electrical and Electronics Engineers Inc., 2021) ALTUNEL, YUSUF; GÜNAYDIN, BURAK; KALABALIKOĞLU, FURKANCurrent man-machine interfaces are far from fitting human expectations in understanding and transmission of noteworthy signs and hints that are natural ingredients of human communication. There are certain indicators and interaction possibilities that can be passed between the two sides but as a result of complex behavior and dynamic changing conditions of environments they can be lost, or at least might require long and complex processing. Gamification is used to enhance the interaction possibilities providing the visual representation of environmental conditions and critical indicators, as well as providing the ability to send and receive requests using suitable techniques for human such as touches, and finger moves on Unity environment. Franchise Management System is selected as a case study and adapted to maintain the interactions between franchiser and franchisee. © 2021 IEEE.Publication Adaptive Direction-Guided Structure Tensor Total Variation(Elsevier, 2021) Kamasak, Mustafa E.; TÜREYEN, EZGİ DEMİRCANDirection-guided structure tensor total variation (DSTV) is a recently proposed regularization term that aims at increasing the sensitivity of the structure tensor total variation (STV) to the changes towards a predetermined direction. Despite of the plausible results obtained on the uni-directional images, the DSTV model is not applicable to the arbitrary (multi-directional and/or partly nondirectional) images. In this study, we build a two-stage denoising framework that brings adaptivity to the DSTV based denoising. We design a DSTV-like alternative to STV, which encodes the first-order information within a local neighborhood under the guidance of spatially varying directional descriptors (i.e., orientation and the dose of anisotropy). In order to estimate those descriptors, we propose an efficient preprocessor that captures the local geometry based on the structure tensor. Through the extensive experiments, we demonstrate how beneficial the involvement of the directional information in STV is, by comparing the proposed method with the state-of-the-art analysis-based denoising models, both in terms of quality and computational efficiency.Publication The Art of Machine Learning as Fashion Stylish for Designing Clothes(Institute of Electrical and Electronics Engineers Inc., 2022) KEYDAL, DUYGU; OYMAK, ERENCAN; DEMİR, KADİR BATUHAN; Yılmaz, Güray; ŞAHİNGÖZ, ÖZGÜR KORAYOver the years, designers have come to the fore with their originality and personal styles and have shaped the fashion industry with their designs. However, due to the progress of time, designers have become unable to meet the demands of all consumers. Since it takes a lot of time to produce an original design, the production process progresses slowly, and customers are uncomfortable with this situation. As in many other industries, designers are trying to solve this problem with the help of artificial intelligence, which is indispensable in the fields of commerce, art, and security. It first entered the fashion sector with drawing programs in the 1950s and has started to change the fashion sector since the 2000s. In the 1950s, artificial intelligence was used only to create a virtual drawing environment When the designer makes a mistake, he can simply erase the mistake and continue working on the design without having to start the whole design from scratch. These programs have greatly facilitated the work of designers. Designers are now able to draw their designs in a much shorter time. But even this shortened period is not enough for the whole fashion industry. Designers could still not keep up with the demands of all customers. Thanks to researchers who added different perspectives to artificial intelligence in the early 2000s with its usage not only for drawing but also for designing, Therefore, in this paper, it is aimed at producing some original designs by preserving the designer's style with the use of Aí techniques. With the proposed model, it has become able to produce ready-made designs by using features such as object detection and visual processing. The experimental results showed that Aí techniques are very successful for combining different patterns for producing an original fashion style. © 2022 IEEE.Publication Biosignal Based Emotion-Oriented Video Summarization(Springer, 2023) DERDİYOK, ŞEYMA; AKBULUT, FATMA PATLARDigital video is a crucial component of multimedia that enhances presentations with accurate, engaging visual and aural data that affects several industries. The transition of video storage from analog to digital is being fueled by a variety of causes. Improved compression methods, cheaper technology, and more network needs are some of these drivers. This paper presents a novel video summarization based on physiological signals provided by emotional stimuli. Through these stimuli, 15 emotions are analyzed using physiological signals. The dataset was gathered from 15 participants who watched 61 episodes of 14 television series while wearing a wristband. We built several deep-learning models for the main purpose of recognizing emotions to summarize video. Among the established networks, the best performance has been obtained with the 1D-CNN, with 92.87% accuracy. This work has been done through a series of empirical experiments; since the frequency of the physiological signals is different, we used models with original and resampled configurations in each experiment. The comprehensive comparison result indicates that the oversampling approach gives the highest accuracy as well as the lowest computational complexity. The performance of the proposed video summarization approach was evaluated by a survey of participants, and the results showed that the summaries contained the critical moments of the video. The proposed approach may be useful and effective in physiological signal-based applications requiring emotion recognition, such as emotion-based video summarization or film genre detection. Additionally, reading such summaries facilitates comprehension of the significance of making rapid judgments regarding likes, ratings, comments, etc.Publication Deep Learning Based Methods for Processing Data in Telemarketing-Success Prediction(Institute of Electrical and Electronics Engineers Inc., 2021) TÜRKMEN, EGEMENIn recent years, the importance of data has been increasing day by day. This has led companies to choose and use them actively, especially for reaching valuable information. Thanks to the interpretation of data, companies can save both time, labor, and costs for these operations in many application areas such as finance, security, e-commerce, data mining, etc. One critical area focuses on the use of finance, in which if the companies properly interpret and use this data, they can directly achieve more successful results in terms of their offering to customers with more accurate campaigns. In this paper, some deep learning methods (Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and Simple Recurrent Neural Network (SimpleRNN)) are used to predict the possibility of subscribing to deposit after the customer is called within the scope of the bank telemarketing campaign. Implemented models are tested with the used dataset and experimental results were compared and interpreted. To improve the obtained accuracy level different approaches are applied to the dataset. Because of the unbalanced structure of the used dataset, SMOTE approach was used to reach more accurate results. After the dataset is processed to be a balanced form, some deep learning methods are applied to it. Obtained results had compared with other proposals. Experimental results showed that the proposed algorithms gave a very acceptable prediction, and it is expected to be used in the finance sector. © 2021 IEEE.Publication Deep Sentiment Analysis With Data Augmentation in Distance Education During the Pandemic(Institute of Electrical and Electronics Engineers Inc., 2022) SOSUN, SERA DENİZ; TAYFUN, BÜLENT; NUKAN, YASEMİN; ALTUN, İREM; ERİK, ELİF BERRA; YILDIRIM, ELİF; KOCAÇINAR, BÜŞRA; AKBULUT, FATMA PATLARDuring the global Covid-19 pandemic, the shutdown of educational institutes has resulted in a phenomenal surge in online learning. Academic activities were shifted to online learning platforms to restrict the influence of COVID-19 and block its spread. For both students and parents, the efficiency of online learning is a major concern, particularly in terms of its suitability for students and teachers, as well as its technological applicability in various social situations. Before the online learning approach can be employed on such a big scale, such challenges must be viewed from different aspects. This study aims to assess the efficiency of online learning by examining individuals' sentiments toward it. Due to social media becoming such an essential form of communication, people's opinions can be observed on platforms like Twitter. The main motivation is to use a Twitter dataset featuring online learning-related tweets. Briefly, we focused on specifying the impact of the Covid-19 pandemic on education in many aspects and parameters by using tweets. We utilized natural language processing models for text classification with a gathered dataset that includes fetching tweets consisting of Covid-19 and education topics. We developed a fine-tuned Long short-term memory (LSTM) model that utilizes data augmentation for classifying the emotional states of individuals. With the deep sentiment analysis model that we proposed, we observed that the negative sentiments were experienced more. © 2022 IEEE.Item Detecting Phishing Websites Using Machine Learning(Institute of Electrical and Electronics Engineers Inc., 2022) ALREFAAI, SAFA; ÖZDEMİR, GHINA; MOHAMED, AFNANPhishing, a cybercriminal's attempted attack, is a social web-engineering attack in which valuable data or personal information might be stolen from either email addresses or websites. There are many methods available to detect phishing, but new ones are being introduced in an attempt to increase detection accuracy and decrease phishing websites ' success to steal information. Phishing is generally detected using Machine Learning methods with different kinds of algorithms. In this study, our aim is to use Machine Learning to detect phishing websites. We used the data from Kaggle consisting of 86 features and 11,430 total URLs, half of them are phishing and half of them are legitimate. We trained our data using Decision Tree (DT), Random Forest (RF), XGBoost, Multilayer Perceptrons, K-Nearest Neighbors, Naive Bayes, AdaBoost, and Gradient Boosting and reached the highest accuracy of 96.6using X G Boost. © 2022 IEEE.Publication Detection of Phishing Websites by Using Machine Learning-Based URL Analysis(Institute of Electrical and Electronics Engineers Inc., 2020) Korkmaz, Mehmet; ŞAHİNGÖZ, ÖZGÜR KORAY; Diri, BanuIn recent years, with the increasing use of mobile devices, there is a growing trend to move almost all real-world operations to the cyberworld. Although this makes easy our daily lives, it also brings many security breaches due to the anonymous structure of the Internet. Used antivirus programs and firewall systems can prevent most of the attacks. However, experienced attackers target on the weakness of the computer users by trying to phish them with bogus webpages. These pages imitate some popular banking, social media, e-commerce, etc. sites to steal some sensitive information such as, user-ids, passwords, bank account, credit card numbers, etc. Phishing detection is a challenging problem, and many different solutions are proposed in the market as a blacklist, rule-based detection, anomaly-based detection, etc. In the literature, it is seen that current works tend on the use of machine learning-based anomaly detection due to its dynamic structure, especially for catching the 'zero-day' attacks. In this paper, we proposed a machine learning-based phishing detection system by using eight different algorithms to analyze the URLs, and three different datasets to compare the results with other works. The experimental results depict that the proposed models have an outstanding performance with a success rate.Publication Genetic Algorithm Based Optimized Waste Collection in Smart Cities(Institute of Electrical and Electronics Engineers Inc., 2020) Özmen, Mehmet; Şahin, Hasan; ŞAHİNGÖZ, ÖZGÜR KORAYIn recent years, the concept of smarts cities emerged to cope with the growth that cities around the world are facing. There are lots of problem areas in smart cities such as smart education, health, buildings, shopping, traffic management, etc. Waste management is one complex and effective problems of urbanization that is needed to be solved in smart cities. Route planning for waste collection and garbage trucks is a known issue in waste management. In this project, a genetic algorithm is proposed to address the problem of waste collection route using a truck fleet. The algorithm was tested in a simplified real state in single area and proved to be applicable to real-world scenarios based solely on the actual data of waste collection of cities.Publication Global Impact of the Pandemic on Education: A Study of Natural Language Processing(Institute of Electrical and Electronics Engineers Inc., 2022) AYAZ, TEOMAN BERKAY; USLU, MUHAMMED SAFA; AĞCABAY, İBRAHİM; AHMED, FARUK; KORKMAZ, ÖMER FARUK; KÜREKSİZ, MESUT; ULUÇAM, EMRE; YILDIRIM, ELİF; KOCAÇINAR, BÜŞRA; AKBULUT, FATMA PATLARSchool closures due to the Covid-19 pandemic have changed education forever and we have witnessed the rise of online learning platforms. The education units of the countries made great efforts to adapt to this new order. The expanding, quick spread of the virus and careful steps have prompted the quest for reasonable choices for continuing education to guarantee students get appropriate education and are not impacted logically or mentally. Different methods were attempted to understand how students were affected by this big change. In addition to the significance of traditional surveys and consulting services, the utilization of social media analysis is used as a supportive approach. This paper analyzes the feedback of students on social media via tweets. Deep sentiment analysis is employed to identify embedded emotions such as negative, neutral, and positive. We also aimed to classify irrelevant tweets as the fourth category. Our experiments showed that the tweets are mostly biased toward negative emotions. © 2022 IEEE.Publication Hybrid Deep Convolutional Model-Based Emotion Recognition Using Multiple Physiological Signals(Taylor & Francis Ltd., 2022) AKBULUT, FATMA PATLAREmotion recognition has become increasingly utilized in the medical, advertising, and military domains. Recognizing the cues of emotion from human behaviors or physiological responses is encouraging for the research community. However, extracting true characteristics from sensor data to understand emotions can be challenging due to the complex nature of these signals. Therefore, advanced feature engineering techniques are required for accurate signal recognition. This study presents a hybrid affective model that employs a transfer learning approach for emotion classification using large-frame sensor signals which employ a genuine dataset of signal fusion gathered from 30 participants using wearable sensor systems interconnected with mobile devices. The proposed approach implements several learning algorithms such as Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), Recurrent Neural Network (RNN), and several other shallow methods on the sensor input to handle the requirements for the traditional feature extraction process. The findings reveal that the use of deep learning methods is satisfactory in affect recognition when a great number of frames is employed, and the proposed hybrid deep model outperforms traditional neural network (overall accuracy of 54%) and deep learning approaches (overall accuracy of 76%), with an average classification accuracy of 93%. This hybrid deep model also has a higher accuracy than our previously proposed statistical autoregressive hidden Markov model (AR-HMM) approach, with 88.6% accuracy. Accuracy assessment was performed by means of several statistics measures (accuracy, precision, recall, F-measure, and RMSE).Publication Intelligent Ambulance Management System in Smart Cities(Institute of Electrical and Electronics Engineers Inc., 2020) Akça, Tugay; ŞAHİNGÖZ, ÖZGÜR KORAY; Koçyiğit, Emre; Tozal, MücahidAccording to the United Nations' expectation, the total population of the cities will be doubled in the next three decades. This accelerating growth causes crucial problems in the main components of both traditional cities and smart cities. To increase the living quality of the residence in smart cities, enabling a clean, healthy, and sustainable environment are the major fields for the smart cities' managers and directors. One of the main infrastructures of the smart city is identified as smart health, which can be enabled with the use of modern technologies such as Internet of Things, especially for accessing the patients when they need help. In this Project, a smart ambulance management system is proposed in a smart city environment. If a patient needs an ambulance, the operator finds the nearest ambulance and direct it to the patient. The coordinates of ambulances are dynamically traced by the system, and Google Maps, as a third-party service, is used in order to calculate the shortest path to the casualty. After reaching to the patient, the expert (doctor or nurse) investigates the situation and finds the best available hospital by the proposed system. The experimental results showed that the proposed system finds the best solution in an acceptable ×.Publication Job Shop Scheduling Problem and Solution Algorithms: A Review(Institute of Electrical and Electronics Engineers Inc., 2020) Çebi, Ceren; Ataç, Enes; ŞAHİNGÖZ, ÖZGÜR KORAYJob Shop Scheduling Problem (JSSP), which aims to schedule several jobs over some machines in which each job has a unique machine route, is one of the NP-hard optimization problems researched over decades for finding optimal sequences over machines. Optimization mainly focused on minimizing the maximum completion time (which is also named as makespan) of whole tasks. According to the size of the problem, JSSP can be defined as Gantt-Chart, Disjunctive Graph, and binary representation forms. This type of scheduling problem is solved with various optimization algorithms such as the Genetic Algorithm, Ant Colony Optimization Algorithm, Particle Swarm Optimization, Tabu Search, or with linear programming models. In this paper, we explain the main characteristics of JSSP and the solution methodologies of this type of problem.Publication Machine Learning Based Heart Disease Detection System(Institute of Electrical and Electronics Engineers Inc., 2022) KIRAR, ARİF TANZERFrom the early days of humanity to today heart diseases take an important place in people's life. The main reason is Heart disease is one of the leading cause of death in the world. The term 'heart disease' refers to several types of heart conditions and it is caused by multiple factors ranging from consumption and daily life style. Since the medical industry began to develop, doctors can detect and cure this diseases more efficiently with the improvement of the technology. In this paper, it is aimed to analyze the relationship between consumed products, our genetic, physical, mental attributes and heart disease and make a model that predicts people have heart disease or not with the help of machine learning and data science. According to experimental results, the proposed approach reached about 95.8 % accuracy for the detection of heart diseases depending on the personal key indicators. © 2022 IEEE.Publication Machine Learning based Human Face Recognition for Attendance System(Institute of Electrical and Electronics Engineers Inc., 2022) HAJJI, KHALED AL; CENADI, ABDULRAHMAN; AHMAD, FAROUQPerson identification and authentication are crucial in business. Many methods are utilized for this identification process. Especially biometric technologies are popular one because of its hardness about deception. Human Face recognition is one of the most preferred tools that helps authenticating the humans. This method allows us to detect changes in a person's face patterns. This technology can be used to identify perpetrators in crime detection and also for getting an accurate attendance systems both in university and also in companies. The part of a person's head from the forehead to the chin, or the corresponding part of an animal, is defined as a face by the Oxford Dictionary. In human. The face is the most crucial aspect in human interactions, since it contains important information about an individual. All humans will acknowledge people from their faces. The proposed solution is to create a working prototype of a system that can help lecturers monitor class attendance in lecture rooms by identifying students' faces from an image taken in the classroom. The database can hold the faces of students once the individual's face matches one of the faces stored in the database, attendance is recorded. Face recognition and detection technologies have been developed in recent years. A number of these are used on social media platforms like Facebook, banking apps, government offices, etc. © 2022 IEEE.Publication Malware Detection in Android Systems with Traditional Machine Learning Models: A Survey(IEEE, 2020) Bayazit, Esra Çalık; ŞAHİNGÖZ, ÖZGÜR KORAY; Doğan, BuketDue to the increased number of mobile devices, they are integrated in every dimension of our daily life. To execute some sophisticated programs, a capable operating must be set up on them. Undoubtedly, Android is the most popular mobile operating system in the world. IT is extensively used both in smartphones and tablets with an open source manner which is distributed with Apache License. Therefore, many mobile application developers focused on these devices and implement their products. In recent years, the popularity of Android devices makes it a desirable target for malicious attackers. Especially sophisticated attackers focused on the implementation of Android malware which can acquire and/or utilize some personal and sensitive data without user consent. It is therefore essential to devise effective techniques to analyze and detect these threats. In this work, we aimed to analyze the algorithms which are used in malware detection and making a comparative analysis of the literature. With this study, it is intended to produce a comprehensive survey resource for the researchers, which aim to work on malware detection.Publication Multi-Modal Fusion Learning Through Biosignal, Audio, and Visual Content for Detection of Mental Stress(Springer London Ltd., 2023) GÜLİN, DOĞAN; AKBULUT, FATMA PATLARMental stress is a significant risk factor for several maladies and can negatively impact a person's quality of life, including their work and personal relationships. Traditional methods of detecting mental stress through interviews and questionnaires may not capture individuals' instantaneous emotional responses. In this study, the method of experience sampling was used to analyze the participants' immediate affective responses, which provides a more comprehensive and dynamic understanding of the participants' experiences. WorkStress3D dataset was compiled using information gathered from 20 participants for three distinct modalities. During an average of one week, 175 h of data containing physiological signals such as BVP, EDA, and body temperature, as well as facial expressions and auditory data, were collected from a single subject. We present a novel fusion model that uses double-early fusion approaches to combine data from multiple modalities. The model's F1 score of 0.94 with a loss of 0.18 is very encouraging, showing that it can accurately identify and classify varying degrees of stress. Furthermore, we investigate the utilization of transfer learning techniques to improve the efficacy of our stress detection system. Despite our efforts, we were unable to attain better results than the fusion model. Transfer learning resulted in an accuracy of 0.93 and a loss of 0.17, illustrating the difficulty of adapting pre-trained models to the task of stress analysis. The results we obtained emphasize the significance of multi-modal fusion in stress detection and the importance of selecting the most suitable model architecture for the given task. The proposed fusion model demonstrates its potential for achieving an accurate and robust classification of stress. This research contributes to the field of stress analysis and contributes to the development of effective models for stress detection.Publication Real-Time Optimization of School Bus Routing Problem in Smart Cities Using Genetic Algorithm(IEEE, 2021) ÖZMEN, MEHMET; ŞAHİN, HASANThe smart city concept, which emerges as a result of the complex structure and major problems of cities, aims to make easy the life of the inhabitant in the urban areas, especially big cities which use IoT devices for the collecting and processing the data in the environment. By using these technologies smart healthcare, smart buildings, smart farming, smart energy planning, smart manufacturing, and smart transporting technologies can be developed. The optimization of school busses is also one of the critical application areas of smart cities that has a direct effect on the students' trip time to schools. In this problem, for collecting the students it is needed to optimize the path of the multiple busses by using different optimization techniques to choose the shortest and the convenient way. In this project, the school busses route problem of a company serving a school was solved with genetic algorithms by getting the related cost values in real-time by using the information from Google Maps. Experimental results showed that, as an evolutionary algorithm, genetic algorithm, can generate acceptable solutions for school busses' routing problems.Publication Sensitivity Reliability Analysis of Power Distribution Networks Using Fuzzy Logic(Institute of Electrical and Electronics Engineers Inc., 2022) Wadi, Mohammed; ELMASRY, WİSAM; Küçük, İsmail; Shahinzadeh, HosseinThis paper proposes a combined method utilizing both the reliability block diagram analytical technique and the Monte Carlo simulation method to estimate the reliability of power systems. Since the reliability of collected data is associated with noise and erroneous data, performing the sensitivity analysis is indispensable. Sensitivity analysis utilizing fuzzy logic specifies these uncertainties and their effects on the reliability calculations. The proposed method is applied to the Roy Billiton Test System Bus-2 to confirm its applicability. The obtained results have verified the sensitivity analysis's importance in drawing an accurate picture of reliability evaluation and a crucial tool for distribution power utilities to identify the susceptible parameters that seriously erode the system's complete reliability. © 2022 IEEE.Publication Three-Way Interaction Model for Turnover Intention of Construction Professionals: Some Evidence from Turkey(ASCE-AMER Soc. Civil Engineers, 2021) UĞURAL, MEHMET NURETTİN; Giritli, HeyecanVoluntary turnover of key employees with scarce skills and tacit knowledge has a potential influence on the competitive advantage of construction organizations. Although much research has been conducted to find out causes of turnover intention, there is limited research in the construction context, particularly considering the influence of the psychological perspective. In this study, using data from 351 construction professionals in the Turkish construction industry and utilizing a three-way moderated moderation model, we examined the interaction effects of psychological factors, operationalized in terms of organizational identification and perceived external prestige, on turnover intention within the boundary of gender differences. Analysis of the responses demonstrates that considering organizational identification, perceived organizational prestige, and gender independently may lead to the underprediction of the turnover intention of employees in construction organizations. The results indicate that organizational identification and perceived organizational prestige have an impact on construction professionals' turnover intention in a condition boundary for gender differences. On this basis, this study may provide important insights for human resource practitioners in identifying potential employees that are likely to have lower turnover intention. It is also of relevance to managers in retaining valued professionals in construction firms.