模糊决策树算法及其在注塑模浇口设计中的知识获取

Fuzzy Decision Tree and Its Knowledge Acquisition in Gate Designing

通过对决策树算法的研究，特别是分析了ＩＤ３的基本算法过程，指出其应用中的多种缺陷，提出了用模糊概念来解决传统决策树中过分匹配，从而提高决策树用于知识获取性能的思想．结合树的深度优先算法、模糊包含度函数和模糊不确定函数，提出了具体实现模糊决策树（ＦＤＴ）算法，并在注塑模浇口设计中加以应用．该算法具有如下优点：在相同精度条件下，大量压缩知识，在将来知识推理中极大地避免了组合爆炸的可能性；随着应用的不断深入，ＦＤＴ获取的知识会逐渐集中在领域中常用的和重要的知识上．

Artificial intelligence can provide a useful path to acquire knowledge automatically. Through research of a decision tree algorithm and its improved methods, especially the basic algorithm of ID3, this article presented their deficiency, such as overfitting in ID3, and gave some resolution ideas with fuzzy concept. Combining the tree's depth first algorithm, fuzzy subsethood and ambiguity measure, a fuzzy decision tree was presented. According to its application in gate designing, we can find two advantages. The first one is that the new algorithm can get smaller and more effective knowledge than the traditional algorithms do. It is very important to avoid knowledge's combination explosion in the knowledge based engineering design system. The other is that the more we use it on the application, the more important and more useful knowledge the algorithm will be focused on.