This paper presents an array pattern synthesis algorithm for arbitrary arrays based on coordinate descent method (CDM). With this algorithm, the complex element weights are found to minimize a weighted L2 norm of the ...This paper presents an array pattern synthesis algorithm for arbitrary arrays based on coordinate descent method (CDM). With this algorithm, the complex element weights are found to minimize a weighted L2 norm of the difference between desired and achieved pattern. Compared with traditional optimization techniques, CDM is easy to implement and efficient to reach the optimum solutions. Main advantage is the flexibility. CDM is suitable for linear and planar array with arbitrary array elements on arbitrary positions. With this method, we can configure arbitrary beam pattern, which gives it the ability to solve variety of beam forming problem, e.g. focused beam, shaped beam, nulls at arbitrary direction and with arbitrary beam width. CDM is applicable for phase-only and amplitude-only arrays as well, and furthermore, it is a suitable method to treat the problem of array with element failures.展开更多
To improve the efficiency and fairness of the spectrum allocation for ground communication assisted by unmanned aerial vehicles(UAVs),a joint optimization method for on-demand deployment and spectrum allocation of UAV...To improve the efficiency and fairness of the spectrum allocation for ground communication assisted by unmanned aerial vehicles(UAVs),a joint optimization method for on-demand deployment and spectrum allocation of UAVs is proposed,which is modeled as a mixed-integer non-convex optimization problem(MINCOP).An algorithm to estimate the minimum number of required UAVs is firstly proposed based on the pre-estimation and simulated annealing.The MINCOP is then decomposed into three sub-problems based on the block coordinate descent method,including the spectrum allocation of UAVs,the association between UAVs and ground users,and the deployment of UAVs.Specifically,the optimal spectrum allocation is derived based on the interference mitigation and channel reuse.The association between UAVs and ground users is optimized based on local iterated optimization.A particle-based optimization algorithm is proposed to resolve the subproblem of the UAVs deployment.Simulation results show that the proposed method could effectively improve the minimum transmission rate of UAVs as well as user fairness of spectrum allocation.展开更多
随着大规模数据的增加,解决Lasso问题成为一个新的热点,以往的方法很难满足大数据背景下的时间和效率问题。为了解决大规模数据及高维数据而带来的计算和储存的困难,本文从三个方面分析最新的算法,即一阶方法、随机方法及并行和分布计...随着大规模数据的增加,解决Lasso问题成为一个新的热点,以往的方法很难满足大数据背景下的时间和效率问题。为了解决大规模数据及高维数据而带来的计算和储存的困难,本文从三个方面分析最新的算法,即一阶方法、随机方法及并行和分布计算。本文介绍和分析了解决最小收缩和选择算子(Least absolute shrinkage and selection operator,Lasso)问题的最新算法:梯度下降方法、交替方向乘子法(Alternating direction method of multipliers,ADMM)和坐标下降方法。其中梯度下降结合一阶方法和Nesterov的加速和光滑技术;交替方向乘子方法将随机方法融入在最新的算法中;坐标下降方法利用其坐标系的特点结合一阶方法、随机方法和并行和分布计算,本文分别从原始目标函数和对偶目标函数的角度对算法进行分析和研究。展开更多
在现实生活中的很多应用里,对不同类别的样本错误地分类往往会造成不同程度的损失,这些损失可以用非均衡代价来刻画.代价敏感学习的目标就是最小化总体代价.提出了一种新的代价敏感分类方法——代价敏感大间隔分布学习机(cost-sensitive...在现实生活中的很多应用里,对不同类别的样本错误地分类往往会造成不同程度的损失,这些损失可以用非均衡代价来刻画.代价敏感学习的目标就是最小化总体代价.提出了一种新的代价敏感分类方法——代价敏感大间隔分布学习机(cost-sensitive large margin distribution machine,CS-LDM).与传统的大间隔学习方法试图最大化"最小间隔"不同,CS-LDM在最小化总体代价的同时致力于对"间隔分布"进行优化,并通过对偶坐标下降方法优化目标函数,以有效地进行代价敏感学习.实验结果表明,CS-LDM的性能显著优于代价敏感支持向量机CS-SVM,平均总体代价下降了24%.展开更多
经典的稀疏表示分类(Sparse Representation for Classification,SRC)算法是一种基于L_1范数最小化问题,它在很多应用场合都能取得很好的分类效果,是目前备受关注的一类识别算法。然而,传统的SRC算法在求解L_1范数最小化问题时,往往计...经典的稀疏表示分类(Sparse Representation for Classification,SRC)算法是一种基于L_1范数最小化问题,它在很多应用场合都能取得很好的分类效果,是目前备受关注的一类识别算法。然而,传统的SRC算法在求解L_1范数最小化问题时,往往计算效率比较低。为有效解决这个问题,提出了一种快速有效的分类算法,它利用坐标下降方法来实现SRC算法。该方法既可以显著地提高计算效率,又可取得较好的分类结果。在不同人脸库上的实验表明,所提的算法具有良好的应用前景。展开更多
文摘This paper presents an array pattern synthesis algorithm for arbitrary arrays based on coordinate descent method (CDM). With this algorithm, the complex element weights are found to minimize a weighted L2 norm of the difference between desired and achieved pattern. Compared with traditional optimization techniques, CDM is easy to implement and efficient to reach the optimum solutions. Main advantage is the flexibility. CDM is suitable for linear and planar array with arbitrary array elements on arbitrary positions. With this method, we can configure arbitrary beam pattern, which gives it the ability to solve variety of beam forming problem, e.g. focused beam, shaped beam, nulls at arbitrary direction and with arbitrary beam width. CDM is applicable for phase-only and amplitude-only arrays as well, and furthermore, it is a suitable method to treat the problem of array with element failures.
基金supported by Project funded by China Postdoctoral Science Foundation(No.2021MD703980)。
文摘To improve the efficiency and fairness of the spectrum allocation for ground communication assisted by unmanned aerial vehicles(UAVs),a joint optimization method for on-demand deployment and spectrum allocation of UAVs is proposed,which is modeled as a mixed-integer non-convex optimization problem(MINCOP).An algorithm to estimate the minimum number of required UAVs is firstly proposed based on the pre-estimation and simulated annealing.The MINCOP is then decomposed into three sub-problems based on the block coordinate descent method,including the spectrum allocation of UAVs,the association between UAVs and ground users,and the deployment of UAVs.Specifically,the optimal spectrum allocation is derived based on the interference mitigation and channel reuse.The association between UAVs and ground users is optimized based on local iterated optimization.A particle-based optimization algorithm is proposed to resolve the subproblem of the UAVs deployment.Simulation results show that the proposed method could effectively improve the minimum transmission rate of UAVs as well as user fairness of spectrum allocation.
文摘随着大规模数据的增加,解决Lasso问题成为一个新的热点,以往的方法很难满足大数据背景下的时间和效率问题。为了解决大规模数据及高维数据而带来的计算和储存的困难,本文从三个方面分析最新的算法,即一阶方法、随机方法及并行和分布计算。本文介绍和分析了解决最小收缩和选择算子(Least absolute shrinkage and selection operator,Lasso)问题的最新算法:梯度下降方法、交替方向乘子法(Alternating direction method of multipliers,ADMM)和坐标下降方法。其中梯度下降结合一阶方法和Nesterov的加速和光滑技术;交替方向乘子方法将随机方法融入在最新的算法中;坐标下降方法利用其坐标系的特点结合一阶方法、随机方法和并行和分布计算,本文分别从原始目标函数和对偶目标函数的角度对算法进行分析和研究。
文摘在现实生活中的很多应用里,对不同类别的样本错误地分类往往会造成不同程度的损失,这些损失可以用非均衡代价来刻画.代价敏感学习的目标就是最小化总体代价.提出了一种新的代价敏感分类方法——代价敏感大间隔分布学习机(cost-sensitive large margin distribution machine,CS-LDM).与传统的大间隔学习方法试图最大化"最小间隔"不同,CS-LDM在最小化总体代价的同时致力于对"间隔分布"进行优化,并通过对偶坐标下降方法优化目标函数,以有效地进行代价敏感学习.实验结果表明,CS-LDM的性能显著优于代价敏感支持向量机CS-SVM,平均总体代价下降了24%.
文摘经典的稀疏表示分类(Sparse Representation for Classification,SRC)算法是一种基于L_1范数最小化问题,它在很多应用场合都能取得很好的分类效果,是目前备受关注的一类识别算法。然而,传统的SRC算法在求解L_1范数最小化问题时,往往计算效率比较低。为有效解决这个问题,提出了一种快速有效的分类算法,它利用坐标下降方法来实现SRC算法。该方法既可以显著地提高计算效率,又可取得较好的分类结果。在不同人脸库上的实验表明,所提的算法具有良好的应用前景。
