Two U.S. patents issued on personalized federated learning

Two U.S. patents from the group were issued on 22 October 2024. US12124963B2 covers disentangled personalized federated learning by consensus representation extraction and diversity propagation. US12124964B2 covers a node-model update that resists bias transfer in federated learning. Inventors include Zhenan Sun, Yunlong Wang, Zhengquan Luo, Kunbo Zhang, Qi Li, and Yong He.