CSE Advances Research Excellence 

2 min read

YEREVAN, Armenia — The American University of Armenia (AUA) Zaven P. & Sonia Akian College of Science and Engineering (CSE) is continually advancing its commitment to research excellence. In this regard, among recent faculty accomplishments are two studies co-authored by CSE Assistant Professor Dr. Varduhi Yeghiazaryan: “Interactive Image Segmentation with Superpixel Propagation” and “Visualization-Driven Graph Sampling Strategy for Exploring Large-Scale Networks.” The co-researchers of the latter study are CSE Adjunct Lecturers Gagik Khalafyan and Irina Tirosyan. Both studies are slated for publication in Springer’s Lecture Notes in Computer Science (LNCS) and a companion volume in its satellite series on Communications in Computer and Information Science (CCIS) within the scope of the proceedings of the 11th International Conference on Analysis of Images, Social Networks and Texts (AIST 2023). 

In the “Interactive Image Segmentation with Superpixel Propagation” study, Dr. Yeghiazaryan focuses on proposing a novel approach to interactive image segmentation. This process separates meaningful regions from inside an image, following user guides and corrections. With its extensive utility in various spheres, this application of artificial intelligence can be applied widely, from photo editing to the analysis of medical and astronomical images. The study uses common open image datasets and an in-house dataset to experiment with high-quality segmentation. In reference to the findings from her study, Dr. Yeghiazaryan explains, “We have proposed a segmentation pipeline that gives the user full control of the process and the results, meanwhile also showing that current state-of-the-art approaches tend to stagnate in quality, despite user input.” By concentrating on the segmentation problem, the study has produced more promising approaches and techniques than many other AI state-of-the-art solutions. 

The “Visualization-Driven Graph Sampling Strategy for Exploring Large-Scale Networks” study delves into current graph sampling techniques and brings a fresh approach to address the shortcomings of existing leading methods. Throughout the study, CSE faculty used the Stanford Large Network Dataset Collection. Hence, its focal point is its concentration on the visual qualities of networks within the system, which differentiates the study from others that use a more theoretical approach. Further, co-researcher Khalafyan elucidates that with the utilization of an improved sampling method, the study demonstrates that the examination of smaller representative segments of large graphs becomes more efficient and easier to explore visually. The sampling method identified for this research was Mino-Centric Graph Sampling (MCGS). 

The approach used in the study has resulted in more accurate sample generation and faster execution times applying the improved “Batch-major Connected Component MCGS” method. Khalafyan remarks: “As datasets continue to grow in complexity and size, having a robust graph sampling strategy becomes even more essential for efficient visualization and analysis. By addressing the limitations of existing sampling techniques, our ‘Batch-major Connected Component MCGS’ method sets a new benchmark in this domain.”

CSE continues to grow its research portfolio consistently striving for strengthening the interconnection between academic and research endeavors. 

Founded in 1991, the American University of Armenia (AUA) is a private, independent university located in Yerevan, Armenia, affiliated with the University of California, and accredited by the WASC Senior College and University Commission in the United States. AUA provides local and international students with Western-style education through top-quality undergraduate and graduate degree and certificate programs, promotes research and innovation, encourages civic engagement and community service, and fosters democratic values.