Dear Editor,Through distributed machine learning,multi-UAV systems can achieve global optimization goals without a centralized server,such as optimal target tracking,by leveraging local calculation and communication w...Dear Editor,Through distributed machine learning,multi-UAV systems can achieve global optimization goals without a centralized server,such as optimal target tracking,by leveraging local calculation and communication with neighbors.In this work,we implement the stochastic gradient descent algorithm(SGD)distributedly to optimize tracking errors based on local state and aggregation of the neighbors'estimation.However,Byzantine agents can mislead neighbors,causing deviations from optimal tracking.We prove that the swarm achieves resilient convergence if aggregated results lie within the normal neighbors'convex hull,which can be guaranteed by the introduced centerpoint-based aggregation rule.In the given simulated scenarios,distributed learning using average,geometric median(GM),and coordinate-wise median(CM)based aggregation rules fail to track the target.Compared to solely using the centerpoint aggregation method,our approach,which combines a pre-filter with the centroid aggregation rule,significantly enhances resilience against Byzantine attacks,achieving faster convergence and smaller tracking errors.展开更多
In order processing in the industrial Internet platform for textile and clothing,assigning optimal order quantities to each factory is the focus and the existing difficulty.The order allocation is a typical NP⁃hard pr...In order processing in the industrial Internet platform for textile and clothing,assigning optimal order quantities to each factory is the focus and the existing difficulty.The order allocation is a typical NP⁃hard problem in combinatorial optimization,and typical research of this kind is still at the initial stage.This paper aims to improve the optimization approach to select factories and to allocate proper orders to each one.It designs a genetic algorithm by making a deviation constraint rule for the initial population and introducing a penalty function to improve convergence.Remarkably,the objective functions of total cost along with the related constraints undergo optimization in the model.The experimental results indicate that the proposed algorithm can effectively solve the model and provide an optimal order allocation for multi⁃factories with less cost and computational duration.展开更多
Static optimization of logical queries is, in substance, to move selections down as far as possible in evaluating logical queries. This paper extends Ullman's RGG (Rule/Goal Graph) and introduces P- graph, with wh...Static optimization of logical queries is, in substance, to move selections down as far as possible in evaluating logical queries. This paper extends Ullman's RGG (Rule/Goal Graph) and introduces P- graph, with which a wide range of recursive logical queries can be statically optimized top-down and evaluated bottom-up, some of which are usually optimized by dynamic approaches. The paper also shows that for some logical queries the complexity of pushing selections down and computing bottom-up is related to the complexity of base relation in the queries.展开更多
基金supported By Guangdong Major Project of Basic and Applied Basic Research(2023B0303000009)Guangdong Basic and Applied Basic Research Foundation(2024A1515030153,2025A1515011587)+1 种基金Project of Department of Education of Guangdong Province(2023ZDZX4046)Shen-zhen Natural Science Fund(Stable Support Plan Program 20231122121608001),Ningbo Municipal Science and Technology Bureau(ZX2024000604).
文摘Dear Editor,Through distributed machine learning,multi-UAV systems can achieve global optimization goals without a centralized server,such as optimal target tracking,by leveraging local calculation and communication with neighbors.In this work,we implement the stochastic gradient descent algorithm(SGD)distributedly to optimize tracking errors based on local state and aggregation of the neighbors'estimation.However,Byzantine agents can mislead neighbors,causing deviations from optimal tracking.We prove that the swarm achieves resilient convergence if aggregated results lie within the normal neighbors'convex hull,which can be guaranteed by the introduced centerpoint-based aggregation rule.In the given simulated scenarios,distributed learning using average,geometric median(GM),and coordinate-wise median(CM)based aggregation rules fail to track the target.Compared to solely using the centerpoint aggregation method,our approach,which combines a pre-filter with the centroid aggregation rule,significantly enhances resilience against Byzantine attacks,achieving faster convergence and smaller tracking errors.
基金Shanghai Foundation for Development of Industrial Internet Innovation,China(No.2019⁃GYHLW⁃004)。
文摘In order processing in the industrial Internet platform for textile and clothing,assigning optimal order quantities to each factory is the focus and the existing difficulty.The order allocation is a typical NP⁃hard problem in combinatorial optimization,and typical research of this kind is still at the initial stage.This paper aims to improve the optimization approach to select factories and to allocate proper orders to each one.It designs a genetic algorithm by making a deviation constraint rule for the initial population and introducing a penalty function to improve convergence.Remarkably,the objective functions of total cost along with the related constraints undergo optimization in the model.The experimental results indicate that the proposed algorithm can effectively solve the model and provide an optimal order allocation for multi⁃factories with less cost and computational duration.
文摘Static optimization of logical queries is, in substance, to move selections down as far as possible in evaluating logical queries. This paper extends Ullman's RGG (Rule/Goal Graph) and introduces P- graph, with which a wide range of recursive logical queries can be statically optimized top-down and evaluated bottom-up, some of which are usually optimized by dynamic approaches. The paper also shows that for some logical queries the complexity of pushing selections down and computing bottom-up is related to the complexity of base relation in the queries.
