根据分层递阶的原则 ,提出一种将粗糙集理论与 BP神经网络相结合的分类算法。该算法分别用粗糙集理论和 BP神经网络处理决策表中的离散属性和连续属性 ,可以避免对象连续属性离散化中产生不确定的情况。同时 ,粗糙集对于决策表噪声比较敏感 ,BP神经网络可以克服这个缺点。最后 ,对 3个公共数据库的测试验证了该分类算法的有效性。 According to the principle of hierarchical and hierarchical, a classification algorithm combining rough set theory with BP neural network is proposed. The algorithm treats the discrete attributes and continuous attributes in the decision table respectively by using rough set theory and BP neural network, so as to avoid the occurrence of uncertainty in discretization of continuous attributes of the object. At the same time, rough set is more sensitive to decision table noise. BP neural network can overcome this shortcoming. Finally, the test of three public databases verified the effectiveness of the classification algorithm.