2024/10/18 by Min Wen, Chengchang Liu, Wen, Min +7
Computer Science · Mathematics · #Advanced Optimization Algorithms Research #Artificial Intelligence (cs.AI) #Distributed #FOS: Computer and information sciences #FOS: Mathematics #Machine Learning (cs.LG) #Matrix Theory and Algorithms #Optimization and Control (math.OC) #Optimization and Variational Analysis #Parallel #and Cluster Computing (cs.DC)
paper · pdf · doi:10.48550/arxiv.2410.14115
openalex publication_date 2024/10/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Bilevel optimization, crucial for hyperparameter tuning, meta-learning and reinforcement learning, remains less explored in the decentralized learning paradigm, such as decentralized federated learning (DFL). Typically, decentralized bilevel methods rely on both gradients and Hessian matrices to approximate hypergradients of upper-level models. However, acquiring and sharing the second-order oracle is compute and communication intensive. % and sharing this information incurs heavy communication overhead. To overcome these challenges, this paper introduces a fully first-order decentralized method for decentralized Bilevel optimization, C2DFB which is both compute- and communicate-efficient. In C2DFB, each learning node optimizes a min-min-max problem to approximate hypergradient by exclusively using gradients information. To reduce the traffic load at the inner-loop of solving the lower-level problem, C2DFB incorporates a lightweight communication protocol for efficiently transmitting compressed residuals of local parameters. % during the inner loops. Rigorous theoretical analysis ensures its convergence % of the algorithm, indicating a first-order oracle calls of O(ε-4). Experiments on hyperparameter tuning and hyper-representation tasks validate the superiority of C2DFB across various typologies and heterogeneous data distributions.