|
July 17, 2024-GTU Office of Press and Public Relations
Scientific evaluation results for TÜBİTAK 1001 Projects are announced, 5 projects from GTU are to be granted support.
The scientific evaluation process of the projects proposed to the Research Support Programs Presidency (ARDEB) for 2024's first term call under the "1001-Scientific and Technological Research Projects Funding Program" by TÜBİTAK (The Scientific and Technological Research Council of Türkiye) has been completed, and the results have been announced. Projects by 5 academics from Gebze Technical University (GTU) have become entitled to receive support.
The academics from GTU and their research projects to be supported under the "1001-Scientific and Technological Research Projects Funding Program" are as follows:
Prof. Didem Gözüpek-Kocaman from GTU Faculty of Engineering, Department of Computer Engineering, has received support for the project titled “Criticality in Packing Coloring of Graphs”.
About the project: Graph theory is a versatile and interdisciplinary research area with wide-ranging applications in various fields such as computer science, telecommunications, industrial engineering, biology, and social sciences. The project will address various packing coloring problems in graphs. The concept of packing coloring has widespread applications, particularly in frequency assignment for wireless networks, where distance requirements play a crucial role. The frequency assignment problem involves assigning frequencies to transmitters in a wireless network. In a broadcasting network, each transmitter is assigned a frequency channel for its transmissions. Considering both civilian and military applications, the increasing demand leads to spectrum scarcity. Therefore, minimizing the frequency range while preventing interference is a fundamental need, and various studies are being conducted in this regard. The packing coloring approach addressed in this project models the requirement that stations assigned the same broadcast frequency must be far enough apart not to interfere with each other. Additionally, in communication networks, evaluating the network's adaptability in the face of node or link failures is essential. The concept of criticality in graphs is used in these evaluations to understand how the parameters of the network change under specific conditions. Therefore, the project focuses on exploring different aspects of criticality in the packing coloring of graphs, aiming to obtain various characterizations and develop recognition algorithms.
Assoc. Prof. F. İnci Özdemir from GTU Faculty of Science, Department of Molecular Biology and Genetics, has received support for the project titled “Redesign of L-Asparaginase Enzyme Derived from Thermophilic Geobacillus Kaustophilus with Mutant-Immobilized Systems for High Acrylamide Removal in Foods”.
About the project: Acrylamide, as one of the chemical contaminants in foods, poses a serious health threat due to its toxic and carcinogenic nature. Acrylamide forms during the frying and baking of carbohydrate-rich foods. Applying the enzyme L-asparaginase before cooking or processing is one of the most effective methods for acrylamide removal. The enzyme must be stable over a wide pH and temperature range and have a high substrate conversion rate and specificity for its application in the food industry. Additionally, the enzyme should be cost-effective and reusable. Previous studies in our laboratory have identified that the thermophilic Geobacillus kaustophilus L-asparaginase type II enzyme has high acrylamide removal performance. The enzyme was immobilized on various supports, increasing its thermal stability and reusability. The project led by Assoc. Prof. Fatma İnci Özdemir, with contributions from three universities in Türkiye, aims to enhance the enzyme's performance through mutation studies for industrial applications using cost-effective and non-toxic immobilization systems. In the project's continuation, the acrylamide removal efficiency of the improved enzymes will be tested in fried potatoes and roasted coffee. The experiments will bring us closer to the industrial use of this enzyme and mark a significant step in developing a high-value, local, non-toxic, and low-cost L-asparaginase.
Asst. Prof. Figen Öztoprak-Topkaya from GTU Faculty of Engineering, Department of Industrial Engineering, has received support for the project titled “Mathematical Optimization with Noisy Function Evaluations: Theory, Techniques for Data-Driven Constraints, and Applications to Renewable Energy System Design”.
About the Project: Most mathematical optimization methods are designed for cases where the objective function and the constraint functions that define the feasible solution set are known in analytical form. However, in many applied problems, it is either not possible to compute these functions and their derivatives with high precision or doing so is computationally expensive. There are only a few studies in the literature on constrained optimization with noisy function evaluations. There is a significant gap in addressing the need for practically efficient, theoretically sound solution methods that are accessible as software libraries for such problems.
On the other hand, recent technological advances in data science and machine learning have made it possible to work with data-driven models more than ever before. A structure frequently encountered in practical applications and directly related to noisy optimization is where the relationships defining the constraint functions cannot be expressed in analytical form or through a computational routine; instead, these relationships must be learned from a dataset. A recently emerging approach is to operate optimization solvers using representative models of such constraints inferred through machine learning techniques. However, this sequential approach—first learning, then optimizing—requires constructing representative models that are valid over the entire domain of the problem. As a result, this often involves working with complex functions, significantly increasing the computational cost of both learning and optimization.
The main thesis of this project is that—rather than developing completely new methods significantly different from standard techniques—preserving the core mechanisms underlying classical solution methods and adapting them appropriately can lead to effective solutions that still yield satisfactory results even under noise. This view has been motivated by numerical results from previous studies and has also been supported by theoretical results obtained on the (relatively simpler) equality-constrained problem structure. It is anticipated that this approach can be extended to handle (noisy) inequality constraints and “data-driven constraints.” For problems with data-driven constraints, this approach aims to produce solution methods that adapt model-based sequential optimization techniques appropriately—methods that locally learn relatively simple model functions and integrate the learning stage directly into the optimization algorithm. In this way, it is expected that the proposed method will outperform the widespread sequential approach in terms of practical performance and speed. The plan is to validate this expectation on a significant real-world application problem. In this context, the project will focus on constraints related to energy storage in the design of renewable energy systems. In the proposed optimization model, battery lifetime will be represented in a data-driven manner.
Upon completion of the project, the objectives are to: develop a general noise-tolerant solution algorithm with theoretical guarantees and an associated solver; extend this solution methodology to handle data-driven constraints integrated with machine learning techniques; and provide insights into the use of these tools and their potential for solving real-world problems.
Asst. Prof. Hüseyin Çimen from GTU Institute of Biotechnology has received support for the project titled “Investigation of the Effects of Resveratrol-Loaded Umbilical Cord Mesenchymal Stem Cell Exosomes on Parkinson's Disease”.
About the project: Parkinson's disease is one of the most common neurodegenerative diseases worldwide. Positive effects of Sirtuin 1 (SIRT1), one of the NAD+-dependent deacetylases, on neurodegenerative diseases have been demonstrated in various studies, including ours. Neuronal cell death can be prevented by regulating mitochondrial quality control systems and oxidative stress. In our project, mitochondrial and cellular changes mediated by SIRT1 in Parkinson's disease, induced by exosomes prepared from umbilical cord mesenchymal stem cells and loaded with resveratrol, will be investigated through molecular and proteomic-based analyses. The results obtained from this study will lead to further evaluation of potential new projects and animal model applications.
Asst. Prof. Meltem Çelen from GTU Institute of Earth and Marine Sciences has received support for the project titled “Investigation of Quantitative Measures To Reduce Nutrient Loads Causing Mucilage Formation in the Sea of Marmara for Current and Climate Change Conditions Using Analytical Modeling and Machine Learning Integration: The Susurluk Basin Case”.
About the project: The project aims to develop a computer-based methodology to reduce nutrient loads reaching the Sea of Marmara from surface waters under current and future conditions and to produce thematic maps as a tool for decision-makers. For this purpose, Analytical Modeling techniques and different Machine learning techniques (ANN, SVM, RF, etc.) will be used. The analytical modeling approach will be used to identify critical sub-regions of the Susurluk Basin in terms of nutrient pollution. Machine learning techniques will prioritize watershed parameters affecting nutrient quality levels classified according to Surface Water Quality Classes (SWQC) and predict future land use changes. In addition to classification methods, Analytical and Machine learning techniques will be applied separately to predict nutrient loads at the sub-watershed level (regression) and test scenarios for reducing these loads. Prof. Salim Öncel from Gebze Technical University, Güleda Engin and Şeref Naci Engin from Yıldız Technical University, Eyüp Selim Köksal from Samsun Ondokuz Mayıs University, and Günay Yıldız Töre from Namık Kemal University will participate as researchers in the project.
|
