Abstract: Concept learning can be formulated as a problem of searching through a predefined space of potential hypothesis the best fits the training examples，in presented several simple learning algorithms（such as，Find-S，List-Then-Eliminate，candidate-Elimination，and so on） which all consider Boolean-valued function，this paper extends Boolean-valued function to[0，1]-valued，it can be used the ideas of fuzzy logic，for every real number in[0，1]，which can discuss all kinds of learning algorithms.It is defined hypothesis is consistent with a set of training examples by fuzzy distance.

Key words: fuzzy set, concept learning, training examples, hypothesis space, [0, 1]-valued

摘要： 概念学习可以形式化为寻找与训练实例最适合的可能假设的预定义空间，已有的多种算法（比如：Find-S、List-Then-Eliminate、candidate-Elimination等等）都是考虑Boolean-值（即{0，1}）函数。用模糊集合的思想，把{0，1}-值函数扩展到[0，1]-值，[0，1]单位区间的每一个实数，都可以用于考虑概念学习算法，而且，可以用模糊距离和贴近度定义假设与训练实例之集的相容性。

关键词: 模糊集合, 概念学习, 训练实例, 假设空间, [0, 1]-值