An Adaptive Asynchronous Online Acceleration Algorithm for Regression Problems with Uneven Update Rates

In this paper, a distributed asynchronous online optimization algorithm based on the Alternating Direction Multiplier Method (ADMM) is proposed for regression problems. The adoption of an asynchronous update mechanism allows nodes to make decisions and update at different times. In particular, an adaptive acceleration method is designed to address the issue of uneven update rates (UUR) in asynchronous mechanisms, which accelerates the convergence of the algorithm. A sublinear regret bound is established for the proposed algorithm, demonstrating its long-term effectiveness. The performance of the proposed method is demonstrated through a simulation experiment.