摘要
讨论了BP神经网络进行R＆D项目中止决策时存在的缺陷,提出了R＆D项目中止决策的区域映射模型,结合教师区域的范围以及像点集合的平面分布特征设计了每个类的判定标准和分类算法。区域映射模型能够克服BP神经网络收敛速度慢、易陷入局部极小点等不足,保证了实际分类准则与训练准则的一致性,具有很高的识别率和比BP神经网络更快的收敛速度,是R＆D项目中止决策的一个有效的模式识别工具。
In the paper,the shortcomings of BP neural network used for R&D project termination decision are discussed and region mapping model for R&D project termination decision is proposed. According to the boundary and distributing in plane,judging criterion and classifying algorithm are designed. Region mapping model can overcome the disadvantages of BP neural network, which has the slow convergence speed and may easily fall into local minimum points. In addition, region mapping model guarantees inherently the consistence of classification rule with the training rule. It has high identification ratio and faster convergence speed than BP neural network. It becomes an effective tool of pattern identification for R&D project termination decision.
出处
《管理工程学报》
CSSCI
2004年第3期69-73,共5页
Journal of Industrial Engineering and Engineering Management
基金
国家自然科学基金资助项目(69974013)
黑龙江省科学技术计划资助项目(K9912)
哈尔滨工业大学跨学科交叉性研究基金资助项目(HIT.MD2001.22)
关键词
R&D项目中止决策
BP神经网络
模式识别
区域映射模型
R&D project termination decision
BP neural network
pattern identification
region mapping model
