The idea of the inverse optimization problem is to adjust the values of the parameters so that the observed feasible solutions are indeed optimal.The modification cost is measured by different norms,such asl1,l2,l∞no...The idea of the inverse optimization problem is to adjust the values of the parameters so that the observed feasible solutions are indeed optimal.The modification cost is measured by different norms,such asl1,l2,l∞norms and the Hamming distance,and the goal is to adjust the parameters as little as possible.In this paper,we consider the inverse maximum flow problem under the combination of the weighted l2 norm and the weighted Hamming distance,i.e.,the modification cost is fixed in a given interval and depends on the modification out of the given interval.We present a combinatorial algorithm which can be finished in O(nm)to solve it due to the minimum cut of the residual network.展开更多
When designing the topology architecture of a large network,or managing and controlling a run network,the battleneck is always changeable with the increase of the network flow,which must be considered. In this paper ,...When designing the topology architecture of a large network,or managing and controlling a run network,the battleneck is always changeable with the increase of the network flow,which must be considered. In this paper ,af-ter analyzing the Ford_Fulkerson algorithm,we point out the relationship between the network min-cutset and thenetwork bottleneck,present an optimal capacity expansion algorithm based on min-cutest ,and take a network instanceto analyze and prove our algorithm in detail. This algorithm can improve the capacity of network effectively and solvethe bottleneck problem of the network.展开更多
基金This research is supported by the Fundamental Research Funds for the Central Universities(No.20720190068)the China Scholarship Council(No.201706315073).
文摘The idea of the inverse optimization problem is to adjust the values of the parameters so that the observed feasible solutions are indeed optimal.The modification cost is measured by different norms,such asl1,l2,l∞norms and the Hamming distance,and the goal is to adjust the parameters as little as possible.In this paper,we consider the inverse maximum flow problem under the combination of the weighted l2 norm and the weighted Hamming distance,i.e.,the modification cost is fixed in a given interval and depends on the modification out of the given interval.We present a combinatorial algorithm which can be finished in O(nm)to solve it due to the minimum cut of the residual network.
文摘针对高斯混合模型(Gaussian mixture model,GMM)参数选取效率较低的问题,提出了一种在基于GMM的轨迹模仿学习表征中综合求解GMM参数估计的方法.该方法基于多中心聚类算法中的最大最小距离算法改进kmeans算法,得到最优初始聚类中心,并基于贝叶斯信息准则(Bayesian information criterion,BIC)通过遗传算法优化求解,同时获取GMM的4个重要参数.该方法通过提高划分初始数据集的效率,在优化初始聚类中心基础上确定混合模型个数,有效地避免了因为初值敏感而导致的局部极值问题.通过多组仿真实验验证了该方法的有效性.
文摘When designing the topology architecture of a large network,or managing and controlling a run network,the battleneck is always changeable with the increase of the network flow,which must be considered. In this paper ,af-ter analyzing the Ford_Fulkerson algorithm,we point out the relationship between the network min-cutset and thenetwork bottleneck,present an optimal capacity expansion algorithm based on min-cutest ,and take a network instanceto analyze and prove our algorithm in detail. This algorithm can improve the capacity of network effectively and solvethe bottleneck problem of the network.
