Sabancı University Students Achieve Great Success in International Publications

Sabancı University Computer Science and Engineering (CS) undergraduate students Mehmet Altunören and Deniz Elbek have achieved significant success with their academic work being published in internationally respected publication platforms.

Mehmet Altunören's single-authored article was published in the journal Transactions on Machine Learning Research. Titled "Combinatorial Capacity of modReLU Complex Networks: VC-Dimension Bounds and Lower Limits," the study examines the statistical capacity of complex-valued neural networks (CVNNs) in the context of VC-dimension. Specifically, it investigates deep networks with modReLU activation, used in applications where both amplitude and phase information are important, and reveals new limits on the combinatorial capacity of these networks.

Deniz Elbek's article, "BLEST: Blazingly Efficient BFS using Tensor Cores," was also accepted for the ACM International Conference on Supercomputing 2026. This study introduces the BLEST library, developed to run the Breadth First Search (BFS) algorithm on modern GPUs. Distinguished by its unique data structure and high-throughput computation approach, this method offers high performance on large-scale real-world graphs, making it one of the most competitive BFS solutions in the field.