基于遗传规划的GSBAR学习算法

GSBAR learning algorithm based on genetic programming

基于相似关系的广义近似推理方法(GSBAR)需要一个能根据案例数据来确定其函数参数的算法,即GSBAR学习算法。提出了基于遗传规划的GSBAR学习算法,论述了GSBAR学习算法的定义、学习算法任务的简化以及学习算法的组成部分和基本步骤。学习算法的运算实例表明:GSBAR学习算法能够以较大概率搜索到合适的函数参数。GSBAR学习算法提供了根据案例数据来确定GSBAR方法中关键函数的能力,从而使得GSBAR方法具有更好的适应性。

The general similarity-based approximate reasoning method (GSBAR) needs a learning algorithm to find its functional parameters based on various cases. This paper describes a GSBAR learning algorithm based on genetic programming. The definition of this algorithm, simplification of the learning task, main components, and steps were presented. An illustrative example shows that the GSBAR learning algorithm most likely finds the optimal result. The function searching ability of the GSBAR learning algorithm makes the GSBAR algorithm more adaptive to a variety of situations.