欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
在Kohonen提出的SOM（self-organization map）神经网络的基础上，通过拓广SOM网络的获胜节点数量，引入惩罚修正因子，改进邻域和连接权函数等方法提出一种新的SOM即SOMDW（SOM with doublewinner）模型。为了验证该模型的有效性，以旅行商问题（traveling salesman problem，TSP）为例对该模型进行检验，得到了满意的结果。另外为了增强SOMDW网络的动态聚类性能，提高解的精确性，还采用禁忌搜索的搜索方法。
英文摘要：
Based on SOM （self-organization map） neural network developed by Kononen, a novel SOM model, namely SOMDW （SOM with double-winner） is proposed by increasing winner nodes and improving some functions such as neighbor function and connection weight function and introducing direction modifying factor in SOM network. In order to validate SOMDW model, an example of TSP （traveling salesman problem） is applied and some satisfying results are also obtained. In addition, in order to enhance the dynamic competition and clustering capability of SOMDW so that some accurate results is Obtained by using SOM, tabu-search method is also applied.
同期刊论文项目
同项目期刊论文
期刊信息
位置：