12 SDSU Georgia students conduct summer research at SDSU in San Diego


One of SDSU Georgia’s priorities is to provide students with ample practical experience, equipping them with skills to become capable professionals in the job market. This summer, SDSU Georgia financed 12 students to conduct research in a sphere of their professional interest alongside professors at SDSU’s main campus.

Vakhtang Donadze and Shota Amashukeli, working with SDSU assistant professor Dr. Baris Aksanli, researched animal activity detection in a natural environment using IoT. “We studied the process of group formation in chimpanzees. A number of chimpanzees will be selected and placed in a new environment - in this case, an island. We want to observe how these chimpanzees interact, socialize and form groups,” Donadze and Amashukeli said. “From our side, the team of Dr. Aksanli is responsible for designing a portable tracker that will be equipped on every chimpanzee. We will be able to know when chimpanzees interact with each other and know the specific chimpanzees who interact with each other. This research will give scientists an idea about how groups and collectives are formed in nature.”

Ana Lomashvili worked with Dr. Calvin Johnson to analyze nuclear many-body wave functions using group theoretical Casimirs. Lomashvili constructed phase-deformed Casimirs to use them in decomposition of different wave functions. “Usually, wave functions are fragmented, but if the same pattern of fragmentation runs through several different states, then we have quasidynamical symmetries. After decomposition of wave functions, I was using the BIGSTICK shell model to test whether I had quasidynamical symmetries in different states,” Lomashvili said. “My five-week session ended with interesting results, and as Dr. Johnson is a truly supportive professor and the research is computational, I will continue my research from Georgia.”

Lana Gaspariani spent the summer working in a computer lab at SDSU on some of Dr. Xiaobai Liu’s object detection system projects with his Ph.D students. Gaspariani helped to find the best model for real-time video data processing - adeptly identifying and framing vehicles in conditions of limited processing memory. “I started helping the team in the beginning of May by preparing a group of videos for further processing. The videos, labeled with vehicles and pedestrians, give computers a base to organize real-time video into labeled streaming. Afterwards, two different teams started using the data for training systems,” Gaspariani said. “In my case, I had to modify the structure of the data I had, and after that, use it as a foundation for training a Yolo v3, Yolo v2 and Yolo v2 tiny object detection algorithms. This process has shown noticeably different outcomes, and in cases where we needed fast, but not ideally precise outcomes, the Yolo v2 tiny algorithm proved to be pretty much enough.”