Distributed Support Vector Machines Over Dynamic Balanced Directed Networks

作     者：Doostmohammadian, Mohammadreza Aghasi, Alireza Charalambous, Themistoklis Khan, Usman A. 

作者机构：Aalto Univ Sch Elect Engn Espoo 02150 Finland Semnan Univ Fac Mech Engn Semnan *** Iran Georgia State Univ Robinson Coll Business Atlanta GA 30309 USA Tufts Univ Dept Elect & Comp Engn Medford MA 02155 USA 

出 版 物：《IEEE CONTROL SYSTEMS LETTERS》 (IEEE Control Syst. Lett.)

年 卷 期：2022年第6卷

页      面：758-763页

核心收录：

基　　金：European Union's Horizon 2020 Research and Innovation Program Academy of Finland 

主　　题：Support vector machines Manganese Heuristic algorithms Distributed databases Radio frequency Switches Signal processing algorithms Support vector machines distributed optimization matrix perturbation theory 

摘      要：In this letter, we consider the binary classification problem via distributed Support Vector Machines (SVMs), where the idea is to train a network of agents, with limited share of data, to cooperatively learn the SVM classifier for the global database. Agents only share processed information regarding the classifier parameters and the gradient of the local loss functions instead of their raw data. In contrast to the existing work, we propose a continuous-time algorithm that incorporates network topology changes in discrete jumps. This hybrid nature allows us to remove chattering that arises because of the discretization of the underlying CT process. We show that the proposed algorithm converges to the SVM classifier over time-varying weight balanced directed graphs by using arguments from the matrix perturbation theory.