The growing interest in addressing minimax optimization problem has been fueled by recent applications in machine learning.Although extensively studied in the convex–concave regime,where a global solution can be effi...The growing interest in addressing minimax optimization problem has been fueled by recent applications in machine learning.Although extensively studied in the convex–concave regime,where a global solution can be efficiently computed,this paper delves into the minimax problem within the nonconvex–concave setup.We propose an alternating gradient projection algorithm with momentum(M-AGP),belonging to single-loop algorithms that not only are easier to implement but also require only the computation of gradient projection updates.We demonstrate that the proposed algorithm identifies an-stationary point of the nonconvex–strongly concave minimax problem in O(ε-2)iterations,representing the best-known rate in the literature.Finally,we utilize two test problems,namely robust nonlinear regression and an image classification problem,to showcase the efficacy of the proposed algorithm.展开更多
In this paper,we propose an alternating proximal gradient algorithm for solving nonsmooth nonconvex-linear minimax problems with coupled linear constraints,which have attracted wide attention in machine learning,signa...In this paper,we propose an alternating proximal gradient algorithm for solving nonsmooth nonconvex-linear minimax problems with coupled linear constraints,which have attracted wide attention in machine learning,signal processing and many other fields in recent years.The iteration complexity of the proposed algorithm is proved to be O(ε-3)to reach anε-stationary point.To our knowledge,this is the first algorithm with iteration complexity guarantee for solving nonsmooth nonconvex-linear minimax problems with coupled linear constraints.展开更多
This study proposes a novel time-synchronization protocol inspired by stochastic gradient algorithms.The clock model of each network node in this synchronizer is configured as a generic adaptive filter where different...This study proposes a novel time-synchronization protocol inspired by stochastic gradient algorithms.The clock model of each network node in this synchronizer is configured as a generic adaptive filter where different stochastic gradient algorithms can be adopted for adaptive clock frequency adjustments.The study analyzes the pairwise synchronization behavior of the protocol and proves the generalized convergence of the synchronization error and clock frequency.A novel closed-form expression is also derived for a generalized asymptotic error variance steady state.Steady and convergence analyses are then presented for the synchronization,with frequency adaptations done using least mean square(LMS),the Newton search,the gradient descent(GraDes),the normalized LMS(N-LMS),and the Sign-Data LMS algorithms.Results obtained from real-time experiments showed a better performance of our protocols as compared to the Average Proportional-Integral Synchronization Protocol(AvgPISync)regarding the impact of quantization error on synchronization accuracy,precision,and convergence time.This generalized approach to time synchronization allows flexibility in selecting a suitable protocol for different wireless sensor network applications.展开更多
针对可移动阵元同时透射和反射可重构智能表面(Movable Elements Based Simultaneous Transmitting and Reflecting Reconfigurable Intelligent Surface,ME-STAR-RIS)辅助抗干扰系统中信道估计开销巨大的问题,提出一种基于深度确定性...针对可移动阵元同时透射和反射可重构智能表面(Movable Elements Based Simultaneous Transmitting and Reflecting Reconfigurable Intelligent Surface,ME-STAR-RIS)辅助抗干扰系统中信道估计开销巨大的问题,提出一种基于深度确定性策略梯度(Deep Deterministic Policy Gradient,DDPG)算法的双尺度协同优化抗干扰传输方法。首先利用统计信道状态信息(Channel State Information,CSI)优化长时阵元位置,再基于优化后的阵元位置估计瞬时CSI,进而优化短时波束成形。为解决高维连续状态空间以及阵元位置、相移系数等连续动作空间带来的优化难题,引入DDPG算法实现动态策略学习。仿真结果表明,所以方法相较于瞬时CSI联合优化方案虽存在约1.5 b/s/Hz的性能损失,但显著降低了信道估计开销。展开更多
The deep deterministic policy gradient(DDPG)algo-rithm is an off-policy method that combines two mainstream reinforcement learning methods based on value iteration and policy iteration.Using the DDPG algorithm,agents ...The deep deterministic policy gradient(DDPG)algo-rithm is an off-policy method that combines two mainstream reinforcement learning methods based on value iteration and policy iteration.Using the DDPG algorithm,agents can explore and summarize the environment to achieve autonomous deci-sions in the continuous state space and action space.In this paper,a cooperative defense with DDPG via swarms of unmanned aerial vehicle(UAV)is developed and validated,which has shown promising practical value in the effect of defending.We solve the sparse rewards problem of reinforcement learning pair in a long-term task by building the reward function of UAV swarms and optimizing the learning process of artificial neural network based on the DDPG algorithm to reduce the vibration in the learning process.The experimental results show that the DDPG algorithm can guide the UAVs swarm to perform the defense task efficiently,meeting the requirements of a UAV swarm for non-centralization,autonomy,and promoting the intelligent development of UAVs swarm as well as the decision-making process.展开更多
With the continuous development of full tensor gradiometer (FTG) measurement techniques, three-dimensional (3D) inversion of FTG data is becoming increasingly used in oil and gas exploration. In the fast processin...With the continuous development of full tensor gradiometer (FTG) measurement techniques, three-dimensional (3D) inversion of FTG data is becoming increasingly used in oil and gas exploration. In the fast processing and interpretation of large-scale high-precision data, the use of the graphics processing unit process unit (GPU) and preconditioning methods are very important in the data inversion. In this paper, an improved preconditioned conjugate gradient algorithm is proposed by combining the symmetric successive over-relaxation (SSOR) technique and the incomplete Choleksy decomposition conjugate gradient algorithm (ICCG). Since preparing the preconditioner requires extra time, a parallel implement based on GPU is proposed. The improved method is then applied in the inversion of noise- contaminated synthetic data to prove its adaptability in the inversion of 3D FTG data. Results show that the parallel SSOR-ICCG algorithm based on NVIDIA Tesla C2050 GPU achieves a speedup of approximately 25 times that of a serial program using a 2.0 GHz Central Processing Unit (CPU). Real airbome gravity-gradiometry data from Vinton salt dome (south- west Louisiana, USA) are also considered. Good results are obtained, which verifies the efficiency and feasibility of the proposed parallel method in fast inversion of 3D FTG data.展开更多
Among all kinds of wavefront control algorithms in adaptive optics systems, the direct gradient wavefront control algorithm is the most widespread and common method. This control algorithm obtains the actuator voltage...Among all kinds of wavefront control algorithms in adaptive optics systems, the direct gradient wavefront control algorithm is the most widespread and common method. This control algorithm obtains the actuator voltages directly from wavefront slopes through pre-measuring the relational matrix between deformable mirror actuators and Hartmann wavefront sensor with perfect real-time characteristic and stability. However, with increasing the number of sub-apertures in wavefront sensor and deformable mirror actuators of adaptive optics systems, the matrix operation in direct gradient algorithm takes too much time, which becomes a major factor influencing control effect of adaptive optics systems. In this paper we apply an iterative wavefront control algorithm to high-resolution adaptive optics systems, in which the voltages of each actuator are obtained through iteration arithmetic, which gains great advantage in calculation and storage. For AO system with thousands of actuators, the computational complexity estimate is about O(n2) ~ O(n3) in direct gradient wavefront control algorithm, while the computational complexity estimate in iterative wavefront control algorithm is about O(n) ~(O(n)3/2), in which n is the number of actuators of AO system. And the more the numbers of sub-apertures and deformable mirror actuators, the more significant advantage the iterative wavefront control algorithm exhibits.展开更多
Based on the thermal stress distribution for functionally gradient material(FGM)plates,a Genetic Algorithm(GA)method for the thermal stresses optimum design of FGM plate with computer technologies is given.The minimum...Based on the thermal stress distribution for functionally gradient material(FGM)plates,a Genetic Algorithm(GA)method for the thermal stresses optimum design of FGM plate with computer technologies is given.The minimum thermal stresses combination distribution for FGM is obtained.展开更多
基金supported by the National Key R&D Program of China(No.2023YFA1011303)the National Natural Science Foundation of China(Nos.11971083 and 11991024)+1 种基金the Team Project of Innovation Leading Talent in Chongqing(No.CQYC20210309536)the Contract System Project of Chongqing Talent Plan(No.cstc2022ycjh-bgzxm0147).
摘要The growing interest in addressing minimax optimization problem has been fueled by recent applications in machine learning.Although extensively studied in the convex–concave regime,where a global solution can be efficiently computed,this paper delves into the minimax problem within the nonconvex–concave setup.We propose an alternating gradient projection algorithm with momentum(M-AGP),belonging to single-loop algorithms that not only are easier to implement but also require only the computation of gradient projection updates.We demonstrate that the proposed algorithm identifies an-stationary point of the nonconvex–strongly concave minimax problem in O(ε-2)iterations,representing the best-known rate in the literature.Finally,we utilize two test problems,namely robust nonlinear regression and an image classification problem,to showcase the efficacy of the proposed algorithm.
基金supported by the National Natural Science Foundation of China(No.12071279).
摘要In this paper,we propose an alternating proximal gradient algorithm for solving nonsmooth nonconvex-linear minimax problems with coupled linear constraints,which have attracted wide attention in machine learning,signal processing and many other fields in recent years.The iteration complexity of the proposed algorithm is proved to be O(ε-3)to reach anε-stationary point.To our knowledge,this is the first algorithm with iteration complexity guarantee for solving nonsmooth nonconvex-linear minimax problems with coupled linear constraints.
基金funded by Universiti Putra Malaysia under a Geran Putra Inisiatif(GPI)research grant with reference to GP-GPI/2023/9762100.
摘要This study proposes a novel time-synchronization protocol inspired by stochastic gradient algorithms.The clock model of each network node in this synchronizer is configured as a generic adaptive filter where different stochastic gradient algorithms can be adopted for adaptive clock frequency adjustments.The study analyzes the pairwise synchronization behavior of the protocol and proves the generalized convergence of the synchronization error and clock frequency.A novel closed-form expression is also derived for a generalized asymptotic error variance steady state.Steady and convergence analyses are then presented for the synchronization,with frequency adaptations done using least mean square(LMS),the Newton search,the gradient descent(GraDes),the normalized LMS(N-LMS),and the Sign-Data LMS algorithms.Results obtained from real-time experiments showed a better performance of our protocols as compared to the Average Proportional-Integral Synchronization Protocol(AvgPISync)regarding the impact of quantization error on synchronization accuracy,precision,and convergence time.This generalized approach to time synchronization allows flexibility in selecting a suitable protocol for different wireless sensor network applications.
摘要针对可移动阵元同时透射和反射可重构智能表面(Movable Elements Based Simultaneous Transmitting and Reflecting Reconfigurable Intelligent Surface,ME-STAR-RIS)辅助抗干扰系统中信道估计开销巨大的问题,提出一种基于深度确定性策略梯度(Deep Deterministic Policy Gradient,DDPG)算法的双尺度协同优化抗干扰传输方法。首先利用统计信道状态信息(Channel State Information,CSI)优化长时阵元位置,再基于优化后的阵元位置估计瞬时CSI,进而优化短时波束成形。为解决高维连续状态空间以及阵元位置、相移系数等连续动作空间带来的优化难题,引入DDPG算法实现动态策略学习。仿真结果表明,所以方法相较于瞬时CSI联合优化方案虽存在约1.5 b/s/Hz的性能损失,但显著降低了信道估计开销。
基金supported by the Key Research and Development Program of Shaanxi(2022GY-089)the Natural Science Basic Research Program of Shaanxi(2022JQ-593).
摘要The deep deterministic policy gradient(DDPG)algo-rithm is an off-policy method that combines two mainstream reinforcement learning methods based on value iteration and policy iteration.Using the DDPG algorithm,agents can explore and summarize the environment to achieve autonomous deci-sions in the continuous state space and action space.In this paper,a cooperative defense with DDPG via swarms of unmanned aerial vehicle(UAV)is developed and validated,which has shown promising practical value in the effect of defending.We solve the sparse rewards problem of reinforcement learning pair in a long-term task by building the reward function of UAV swarms and optimizing the learning process of artificial neural network based on the DDPG algorithm to reduce the vibration in the learning process.The experimental results show that the DDPG algorithm can guide the UAVs swarm to perform the defense task efficiently,meeting the requirements of a UAV swarm for non-centralization,autonomy,and promoting the intelligent development of UAVs swarm as well as the decision-making process.
基金the Sub-project of National Science and Technology Major Project of China(No.2016ZX05027-002-003)the National Natural Science Foundation of China(No.41404089)+1 种基金the State Key Program of National Natural Science of China(No.41430322)the National Basic Research Program of China(973 Program)(No.2015CB45300)
摘要With the continuous development of full tensor gradiometer (FTG) measurement techniques, three-dimensional (3D) inversion of FTG data is becoming increasingly used in oil and gas exploration. In the fast processing and interpretation of large-scale high-precision data, the use of the graphics processing unit process unit (GPU) and preconditioning methods are very important in the data inversion. In this paper, an improved preconditioned conjugate gradient algorithm is proposed by combining the symmetric successive over-relaxation (SSOR) technique and the incomplete Choleksy decomposition conjugate gradient algorithm (ICCG). Since preparing the preconditioner requires extra time, a parallel implement based on GPU is proposed. The improved method is then applied in the inversion of noise- contaminated synthetic data to prove its adaptability in the inversion of 3D FTG data. Results show that the parallel SSOR-ICCG algorithm based on NVIDIA Tesla C2050 GPU achieves a speedup of approximately 25 times that of a serial program using a 2.0 GHz Central Processing Unit (CPU). Real airbome gravity-gradiometry data from Vinton salt dome (south- west Louisiana, USA) are also considered. Good results are obtained, which verifies the efficiency and feasibility of the proposed parallel method in fast inversion of 3D FTG data.
基金supported by the National Key Scientific and Research Equipment Development Project of China(Grant No.ZDYZ2013-2)the National Natural Science Foundation of China(Grant No.11173008)the Sichuan Provincial Outstanding Youth Academic Technology Leaders Program,China(Grant No.2012JQ0012)
摘要Among all kinds of wavefront control algorithms in adaptive optics systems, the direct gradient wavefront control algorithm is the most widespread and common method. This control algorithm obtains the actuator voltages directly from wavefront slopes through pre-measuring the relational matrix between deformable mirror actuators and Hartmann wavefront sensor with perfect real-time characteristic and stability. However, with increasing the number of sub-apertures in wavefront sensor and deformable mirror actuators of adaptive optics systems, the matrix operation in direct gradient algorithm takes too much time, which becomes a major factor influencing control effect of adaptive optics systems. In this paper we apply an iterative wavefront control algorithm to high-resolution adaptive optics systems, in which the voltages of each actuator are obtained through iteration arithmetic, which gains great advantage in calculation and storage. For AO system with thousands of actuators, the computational complexity estimate is about O(n2) ~ O(n3) in direct gradient wavefront control algorithm, while the computational complexity estimate in iterative wavefront control algorithm is about O(n) ~(O(n)3/2), in which n is the number of actuators of AO system. And the more the numbers of sub-apertures and deformable mirror actuators, the more significant advantage the iterative wavefront control algorithm exhibits.
摘要Based on the thermal stress distribution for functionally gradient material(FGM)plates,a Genetic Algorithm(GA)method for the thermal stresses optimum design of FGM plate with computer technologies is given.The minimum thermal stresses combination distribution for FGM is obtained.