首页> 外国专利> NETWORK TYPE INFORMATION PROCESSING SYSTEM USING GENETIC ALGORITHM, LEARNING DEVICE AND LEARNING METHOD FOR THE SAME, AND RECORDING MEDIUM WITH PROGRAM OF LEARNING METHOD RECORDED

NETWORK TYPE INFORMATION PROCESSING SYSTEM USING GENETIC ALGORITHM, LEARNING DEVICE AND LEARNING METHOD FOR THE SAME, AND RECORDING MEDIUM WITH PROGRAM OF LEARNING METHOD RECORDED

机译：使用遗传算法，学习装置和学习方法的网络类型信息处理系统，以及记录有学习方法的程序的介质

页面导航

摘要
著录项
相似文献

摘要

PPROBLEM TO BE SOLVED: To automate efficient learning and automate input selection by using a genetic algorithm in a network type information processing system and dynamically changing learning parameters in dependence on a learning state. PSOLUTION: A plurality of network type information processing systems are prepared. Learning parameters in each network type information processing system are set as GA parameters. An inference mechanism and the GA parameters in each network type information processing system are combined into an individual, which constitutes a population. On the population, a genetic algorithm including learning, evaluation inference, performance evaluation, and alternation of generations using the GA parameters is executed to select a population of high performance. PCOPYRIGHT: (C)2004,JPO&NCIPI

机译：

要解决的问题：通过在网络类型信息处理系统中使用遗传算法并根据学习状态动态更改学习参数，来实现高效学习的自动化和输入选择的自动化。

解决方案：准备了多个网络类型的信息处理系统。将每个网络类型信息处理系统中的学习参数设置为GA参数。每个网络类型信息处理系统中的推理机制和GA参数都组合成一个个体，构成一个总体。在种群上，执行遗传算法，包括学习，评估推断，性能评估和使用GA参数的世代交替，以选择高性能种群。

版权：（C）2004，日本特许厅和日本国家唱片公司展开▼

著录项

公开/公告号JP2004192584A

专利类型
公开/公告日2004-07-08

原文格式PDF
申请/专利权人 ADOIN KENKYUSHO:KK;
展开▼

申请/专利号JP20020363168
发明设计人 NAKAMURA TAKESHI;SHIOZAWA KEIKO;
展开▼

申请日2002-12-13
分类号G06N3/12;G06N3/00;G06N3/04;
国家 JP
入库时间 2022-08-21 23:27:01

相似文献

专利
外文文献
中文文献