A new heuristic search method is devised for maximum likelihood method , which first constructs a starting tree , and then through branch swapping searches for trees best resembling the present optimal tree and locates a better tree . repeat this process , till no more better trees are generated or the upper iteration limit is arrived and the program ends 对最大似然法,改进了一种新的启发式搜索:随机建立一棵初始树,然后通过查看与当前最优树拓扑最近的树,从中寻找更优树,不断地更新当前最优树,直到无法找到更优树或者达到了搜索次数的上限,算法停止。
Some existing topology - unaware collective operations algorithms do n ' t make use of the information about network topology . firstly , these approaches construct a model of network topology and make use of a set of parameters to describe the communication characteristics of a network model , then build the optimal trees based parameter values and their models to realize collective communication 现有的一些未知拓扑型集合通信算法,并不利用网络拓扑结构信息,都是首先对网络模型化,并且用各种参数来描述该模型中网络通信的特点,然后基于此模型和其参数值为集合通信操作构造最优树来实现集合通信。
Classification and regression trees processing part introduces growing algorithm of cart , pruning algorithm of cart and selecting best tree algorithm etc . on the basis of the concerned new model , the thesis presents in details the designing of multi - rules neural network based cart system for abnormal customers recognition 在分类回归树部分,介绍了分类回归树的生长算法、最小代价?复杂性剪枝算法以及最优树选择等算法。提出了系统设计之后,论文详细介绍了该系统的开发,用以解决异动客户的识别问题。