单一目标概念的学习算法研究

RESEARCH ON LEARNING ARITHMETIC OF SINGLE CONCEPT

归纳学习模型可以用示例空间和假设空间的双空间模型表示 ,描述空间是假设空间中适合已观察示例的假设所构成的集合 ,普遍边界集和特殊边界集约束着整个描述空间 ,单一目标概念学习算法是通过增加已分类示例来更新边界集 ,缩小描述空间 ,最后穷尽描述空间 ,从而得到目标概念。

Inductive learning model can be described by a dual space model,instance space and hypothesis space.The version space is a set of hypotheses which fit observed instances.The version space is bounded by general boundary set and specific boundary set.The learning arithmetic of single concept updates the boundary sets with additional instances,and diminishes the version space.Target concept can be got by exhausting the version space.