Our Computer Engineering student has been awarded funding under the TÜBİTAK 2209-A Research Project Support Program

A research project led by Mehmetcan Mutlu, a senior student in the Department of Computer Engineering (English Program) at our university, and supervised by Prof. Dr. Meltem Huri Baturay, has been awarded funding under the TÜBİTAK 2209-A University Student Research Projects Support Program.

The project aims to develop an integrated multi-task learning architecture that simultaneously calculates the key credit risk parameters (Probability of Default, Loss Given Default, and Risk at Default)—which are calculated in accordance with the TFRS 9 standards, a legal requirement in the banking sector—rather than estimating these parameters using traditional, independent models. This innovative approach utilizes a deep neural network (MTL-NN)-based infrastructure that analyzes the latent correlations among risk parameters.

Based on the results obtained, the plan is to address the discrepancies and incomplete risk measurements caused by traditional approaches that treat risk parameters independently. Thanks to this integrated machine learning model, significantly higher prediction accuracy and consistent risk forecasting can be achieved compared to traditional statistical methods and standalone models.

Ultimately, it is expected that this innovative infrastructure—which successfully integrates deep learning, big data analytics, and financial technologies—will contribute to banks managing their capital adequacy more accurately and provide a modern and innovative direction to our national risk management ecosystem in line with the "Financial Stability" objectives of the 12th Development Plan.