Master Degree / Yüksek Lisans Tezleri

Permanent URI for this collectionhttps://hdl.handle.net/11147/3008

Browse

Filters

2019 - 201912020 - 20234

Settings

Scopus Q: search.filters.scopusQuartile.N/ASubject: search.filters.subject.Neural networks

Search Results

Now showing 1 - 5 of 5
  • Master Thesis
    Recognition of Counterfactual Statements in Turkish
    (01. Izmir Institute of Technology, 2023) Acar, Ali; Tekir, Selma
    Counterfactual statements describe an event that did not happen or cannot happen, and optionally the consequence of this event if it would happen. Counterfactual statements are the building blocks of human thought processes as people constantly reflect upon past happenings and consider their future implications. Counterfactual reasoning is essential for machine intelligence and explainable artificial intelligence studies. Detecting counterfactuals automatically with machine learning algorithms is very crucial for these areas. This thesis presents the development of the first-ever Turkish counterfactual detection dataset. It presents a comprehensive classification baseline and expands the scope of counterfactual detection to include the Turkish language.
  • Master Thesis
    Improvement on Motion-Guided Siamese Object Tracking Networks Using Prioritized Windows
    (01. Izmir Institute of Technology, 2021) Ünlü, Ünver Can; Baştanlar, Yalın
    In recent years, there has been significant progress in Visual Object Tracking with evolutions of both computers and learning algorithms, especially in Neural Networks. Therefore, we obtain better results by combining Neural Networks and traditional tracking methods such as Kalman Filter and Correlation Filters. SiamFC is an example of such algorithms because SiamFC combines Siamese Neural Networks and Correlation Filters. SiamFC is open to development because it does not have an online learning process. An example of the improved SiamFC is Kalman-Siam that combines Kalman Filter and Multi-feature SiamFC. Kalman-Siam uses Kalman-Filter to solve the occlusion situation problem by processing the target's previous motion trajectory. Therefore, the tracking can fail in other complex scenarios for Kalman-Siam. One of the methods for solving such problems is detecting this situation and starting the re-tracking process as we used in this research. Also, we used a parameter calculated on the response map after the correlation operation in SiamFC to detect these situations. First, our algorithm generates possible prioritized search windows. Then, it runs in a specific order of priority for these generated search windows surrounding the target's last known location. We named this process Adaptive Window Search that starts from the highest priority search windows and continues until the lowest search windows do not exist. Therefore, we named our algorithm Adaptive-Kalman-Siam. We demonstrated more successful results on commonly used datasets. Adaptive-Kalman-Siam tracks an object better than SiamFC and Kalman-Siam in Background Clutters, Fast Motion, Motion Blur, and Occlusion complex tracking scenarios.
  • Master Thesis
    Synthetic Generation of Fingerprints
    (Izmir Institute of Technology, 2020) İrtem, Emre; Erdoğmuş, Nesli
    Fingerprints are unique to each person and they have been widely used and accepted for identification purposes by the society. Fingerprints can be captured by using ink and paper to get a print and then digitizing it or more recently by using specialized sensors. But in both cases, trained specialist supervision is mostly needed. Moreover, since fingerprints are personal information, they are protected by the laws on personal data protection. Therefore, collection/sharing of real fingerprints is difficult and illegal without the consent of their owner. On the otherhand, deep learning systems that are proven to be very successfull in many machine learning task, usually depend on very large training sets to achive high accuracies. In this study, to overcome the data hunger problem for training deep neural networks, synthetic fingerprints are generated by using model-based methods. For this purpose, firstly master fingerprint images are generated and next many impressions are derived from them by applying real-world degradations. The realism and the usability of synthetic fingerprints are tried and validated using a fingerprint classification system. For which, a deep neural networks are trained with and without the synthetically generated data. As a result of the experiments, it is shown that the generated fingerprint images are realistic enough to positively effect the classification results and that the usage of the synthetically generated fingerprints in training deep systems are promising.
  • Master Thesis
    A Language Modeling Approach To Detect Bias
    (Izmir Institute of Technology, 2020) Atik, Ceren; Tekir, Selma
    Technology is developing day by day and is involved in every area of our lives. Technological innovations such as artificial intelligence can strengthen social biases that already exist in society, regardless of the developers' intentions. Therefore, researchers should be aware of this ethical issue. In this thesis, the effect of gender bias, which is one of the social biases, on occupation classification is investigated. For this, a new dataset was created by collecting obituaries from the New York Times website and they were handled in two different versions, with and without gender indicators. Since occupation and gender are independent variables, gender indicators should not have an impact on the occupation prediction of models. In this context, in order to investigate gender bias on occupation estimation, a model in which occupation and gender are learned together is evaluated as well as models that make only occupation classification are evaluated. The results obtained from models state that gender bias has a role in classification occupation.
  • Master Thesis
    Robustness of Fingerprint Verification Algorithms Against Synthetic Deformations
    (Izmir Institute of Technology, 2019) Cantürk, Sinem; Erdoğmuş, Nesli
    Fingerprint recognition is one of the biometric techniques used for the identification of humans. The developments and research about fingerprint recognition to date are of great importance in advancing fingerprint recognition and verification scenarios. The fact that fingerprint recognition systems are used almost everywhere and are easily accessible is directly proportionate to a large amount of research in these areas. During the acquisition of the fingerprint, there are many environmental factors that may affect the quality of the print and eventually, its ability to be recognized. For a fingerprint recognition algorithms, it is important to handle the difficulties that arise due to those variations. The aim of the thesis is to obtain and compare the results of not only an existing feature-based fingerprint recognition techniques but a fingerprint recognition technique that uses deep learning. The main focus is on how fingerprint verification algorithms behave under the circumstances of synthetically distorted fingerprint images. After developing two different verification systems, the goal is to compare system results with and without distorted images. The results of the two methods with and without externally added deformations effect on the fingerprint image is compared. The first system has a feature-based approach comparing the images via local features on the fingerprint. In order to do this two different descriptors that are called ORB and SIFT are used. In the feature-based approach, there is also a matching part and this part is tried with two different matching algorithms that are called Brute Force Matcher and Approximate Nearest Neighbor (ANN) matcher. The second algorithm makes the decision of match or non-match by feeding the raw fingerprint images as an input to a deep neural network and comparing the feature vectors calculated by the network. This study has revealed that deep neural network approach has given more robust and faster results on both the original dataset and distorted versions of the dataset.