论文部分内容阅读
BP神经网络方法在知识转移评价研究领域的运用普遍存在样本数量偏少、样本质量和代表性良莠不齐的问题。针对这一问题,本文在自建评价指标体系的基础上采用BP神经网络方法构建综合评价模型,通过大量发放企业问卷进行样本数据采集,用于神经网络的训练、检验及仿真。结果表明,BP神经网络评价模型更接近实际评价过程,能有效规避了常规综合评价方法中人为因素干扰及权重确定的主观性等弊端,从而具有较高的准确性和实用性。
The application of BP neural network in the field of knowledge transfer evaluation research has the problem that the number of samples is generally small, the quality and the representativeness of samples are different. In response to this problem, this paper builds a comprehensive evaluation model based on self-built evaluation index system by using BP neural network method, collects sample data by issuing a large number of enterprises questionnaire, and is used for training, testing and simulation of neural network. The results show that the BP neural network evaluation model is closer to the actual evaluation process, which can effectively avoid the shortcomings of the subjectiveness of human factors interference and weight determination in the conventional comprehensive evaluation method, which has higher accuracy and practicability.