<ul><li>A novel framework is proposed for heterogeneous federated learning (FL) to address client heterogeneity and improve model performance.</li><li>The framework captures local and global training processes through a bilevel formulation.</li><li>It includes personalized learning, pre-training on the server's side, nonstandard aggregation, nonidentical local steps, and clients' local constraints.</li><li>The proposed method, ZO-HFL, achieves nonasymptotic and asymptotic convergence guarantees without relying on standard assumptions in heterogeneous FL.</li></ul>

A Randomized Zeroth-Order Hierarchical Framework for Heterogeneous Federated Learning

Discover more