期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
集成特征选择算法和LightGBM融合的分类模型 被引量:2
1
作者 王选 刘祥伟 《福建电脑》 2022年第4期12-15,共4页
实体零售网点具有数据特征种类多、数据结构复杂的特点。为了对实体零售网点进行分类分级管理,本文采用改进的集成特征选择算法ISFA和LightGBM算法对零售网点进行数据特征优化选择和分类分级。首先采用集成特征选择方法对零售网点的特... 实体零售网点具有数据特征种类多、数据结构复杂的特点。为了对实体零售网点进行分类分级管理,本文采用改进的集成特征选择算法ISFA和LightGBM算法对零售网点进行数据特征优化选择和分类分级。首先采用集成特征选择方法对零售网点的特征进行筛选,然后用LightGBM算法对筛选出的特征子集进行分类和预测。以网点基本数据和统计数据为对象进行对比实验。结果表明,对于不同网点的分类目标,使用本文的方法可取得良好的效果。 展开更多
关键词 零售网点 网点分类 特征选择 LightGBM
下载PDF
On-demand multicast routing protocol based on node classification in MANET 被引量:2
2
作者 邓霞 孙利民 +2 位作者 王建新 罗玉宏 陈建二 《Journal of Central South University of Technology》 EI 2006年第2期190-195,共6页
An improved on-demand multicast routing protocol(ODMRP), node classification on-demand multicast routing protocol(NC-ODMRP), which is based on node classification in mobile ad hoc networks was proposed. NC-ODMRP class... An improved on-demand multicast routing protocol(ODMRP), node classification on-demand multicast routing protocol(NC-ODMRP), which is based on node classification in mobile ad hoc networks was proposed. NC-ODMRP classifies nodes into such three categories as ordinary node, forwarding group(FG) node, neighbor node of FG node according to their history forwarding information. The categories are distinguished with different weights by a weight table in the nodes. NC-ODMRP chooses the node with the highest weight as an FG node during the setup of forwarding group, which reduces a lot of redundant FG nodes by sharing more FG nodes between different sender and receiver pairs. The simulation results show that NC-ODMRP can reduce more than 20% FG number of ODMRP, thus enhances nearly 14% data forwarding efficiency and 12% energy consumption efficiency when the number of multicast senders is more than 5. 展开更多
关键词 mobile ad hoc networks MULTICAST forwarding group(FG) node classification
下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部