Jae-Won Yoon (Integrated PhD Program) and Yong-Jung Oh (PhD Candidate) Win “Outstanding Paper Award” at the 2023 KICS Summer Conference

Jae-Won Yoon, an integrated PhD program student, and Yong-Jung Oh, a PhD candidate from POSTECH’s Department of Electrical Engineering (advisor: Professor Yosep Jeon), were awarded the Outstanding Paper Award at the 2023 Summer Conference of the Korean Institute of Communications and Information Sciences (KICS). Their paper, titled “Sparsification-Based Compression Techniques for Communication-Efficient Federated Learning,” was recognized for its innovative contribution to the field.

Research Highlights

The study addresses the communication overhead problem in federated learning by proposing a novel compression method using sparsification and quantization for model updates. Additionally, it introduces recovery techniques leveraging the Bussgang method and Linear Minimum Mean Squared Error (LMMSE) detection.

The results demonstrate that the proposed compression and recovery methods significantly reduce communication overhead while maintaining high performance in federated learning systems. This breakthrough has the potential to enhance the scalability and efficiency of federated learning in real-world applications.