Today marks an important milestone for Ruslan Zhagypar, a master's student in Electrical and Computer Engineering, as he successfully defended his thesis on "Characterization of the Global Bias Problem in Aerial Federated Learning." The master's defense, held at King Abdullah University of Science and Technology (KAUST) in the Information Science Lab (ISL), showcased Ruslan's extensive research in the field of aerial federated learning (FL) and its associated challenges.
Ruslan's thesis focused on addressing the global bias problem in aerial FL. With the increasing use of unmanned aerial vehicles (UAVs) in data collection, FL has emerged as a promising approach for training models on distributed UAV networks while keeping data privacy. However, global bias introduced as a result of unreliable wireless channel can significantly affect the accuracy and performance of the FL algorithms.
During his research, Ruslan explored and characterized the global bias problem in aerial FL using stochastic geometry tools. This allowed him to devise an approach to identify, quantify, and mitigate biases within the FL framework. By examining the impact of biases on model performance, Ruslan's work contributes to enhancing the reliability and fairness of machine learning models trained on distributed UAV networks.
Ruslan's remarkable achievement would not have been possible without the guidance and support of his supervisor, Prof. Tareq Y. Al-Naffouri, and the valuable contributions of Dr. Nour Kouzayha, Dr. Hesham ElSawy, and Dr. Hayssam Dahrouj. They provided invaluable support as senior research staff members, contributing their expertise to enhance the quality of Ruslan's work. We congratulate Ruslan for his accomplishment, and we look forward to his continued success in his future endeavors.
