Abstract: This paper considers the distributed online optimization (DOO) problem over time-varying unbalanced networks, where gradient information is explicitly unknown. To address this issue, a ...