基于FOIL算法的约束归纳方法研究

Constraint Inductive Research Based on FOIL Algorithm

对FOIL算法进行了深入剖析,针对归纳逻辑程序设计处理数值量弱的缺点,在FOIL算法的基础上作了改进。研究一种新的约束归纳算法,该算法通过在特殊化时对假设空间的划分和引入Fisher判别方法,能够导出不受变量个数限制的多种形式的线性约束。实验结果证明,改造后的算法在约束求解方面具有有效性。

After a thorough Analysis of the essence of foil algorithm, considering the weakness of ILP＇s ability to the numerical value, this paper make some improvement on the foil algorithm. And proposes a new approach to constraint inductive logic programming. This new approach generate various forms of liner constraint with multiple variables by the way of partitioning it＇s hypothesis space and introducing Fisher. In the end, an example is shown to prove this new approach＇s efficiency.