学习系统的重学习收敛性能分析  

Convergence performance analysis of multi-learning of learning system.

作　　者：董苗波[1] 吴铁军[1] 

机构地区：[1]浙江大学智能系统与决策研究所

出　　处：《浙江大学学报（理学版）》2002年第2期213-218,共6页Journal of Zhejiang University（Science Edition）

摘　　要：重学习是指从不只一个初始点开始的学习 ,本文分析了学习系统的重学习收敛性能 ,给出了描述重学习的结果优劣和初始点集稠密程度之间关系的定理 .并对一个优化问题引出的学习系统做了重学习的仿真 。Multi learning is learning from more than one point called an initial point set. The convergence performance of multi learning is analysed, and a theorem describing the relationship between the qualification of the multi learning result and its initial points set's density is proved. As an example, an optimization problem is brought forward as a learning system and multi learnings of which from several initial points sets of different densities are simulated. The results demonstrate the relationship proposed by the theorem.

关 键 词：学习系统 重学习 收敛性能 学习能力 初始点稠密程度 学习结果优化程度 

分 类 号：TP273.22[自动化与计算机技术—检测技术与自动化装置]