One-Point Sampling for Distributed Bandit Convex Optimization with Time-Varying Constraints
One-Point Sampling for Distributed Bandit Convex Optimization with Time-Varying Constraints
This paper considers the distributed bandit convex optimization problem with time-varying constraints. In this problem, the global loss function is the average of all the local convex loss functions, which are unknown beforehand. Each agent iteratively makes its own decision subject to time-varying inequality constraints which can be violated but are fulfilled in the long run. For a uniformly jointly strongly connected time-varying directed graph, a distributed bandit online primal-dual projection algorithm with one-point sampling is proposed. We show that sublinear dynamic network regret and network cumulative constraint violation are achieved if the path-length of the benchmark also increases in a sublinear manner. In addition, an $\mathcal{O}({T^{3/4 + g}})$ static network regret bound and an $\mathcal{O}( {{T^{1 - {g}/2}}} )$ network cumulative constraint violation bound are established, where $T$ is the total number of iterations and $g \in ( {0,1/4} )$ is a trade-off parameter. Moreover, a reduced static network regret bound $\mathcal{O}( {{T^{2/3 + 4g /3}}} )$ is established for strongly convex local loss functions. Finally, a numerical example is presented to validate the theoretical results.
Xinlei Yi、Guanghui Wen、Lihua Xie、Tianyou Chai、Tao Yang、Kunpeng Zhang、Lei Xu
计算技术、计算机技术
Xinlei Yi,Guanghui Wen,Lihua Xie,Tianyou Chai,Tao Yang,Kunpeng Zhang,Lei Xu.One-Point Sampling for Distributed Bandit Convex Optimization with Time-Varying Constraints[EB/OL].(2025-04-22)[2025-05-12].https://arxiv.org/abs/2504.16211.点此复制
评论