中文摘要：

传统模糊积分将高维数据投影到一维空间进行分类和预测，但很多时候无法全面描述和正确分类。提出一种深度模糊积分，由双重模糊积分延伸至多重模糊积分，以适应复杂的数据分布，提高分类精度。当数据被投影至一维空间效果不佳时，进行第二次投影，两次投影所得的值构成二维坐标，也就是将投影至一维空间的点拉伸至二维空间进行分类，依此类推，可扩展至多维空间，扩展后的维度可根据复杂度情况学习确定。实验分为两部分，分别是应用于经典数据库和关于肝炎疾病的真实数据。结果表明，深度模糊积分与几种经典算法比较，分类精确度都有不同程度的提高。由此可见，深度模糊积分对经典模糊积分进行了纵深扩展，解决了数据交叠问题，提高了分类性能。

英文摘要：

The traditional fuzzy integral is applied to classification or prediction by projecting data from high dimensional space to one dimensional space. This paper proposed a generalized fuzzy measure called deep fuzzy integral, which could be accustomed to complex data distribution and promote the classification accuracy. When the results were not satisfied with projection to 1 dimension space, it executed another projection to stretch the data in 1 dimension to multiple dimensions. This paper performed experiments on two parts which included classical datasets and HBV data. The results show that the deep fuzzy integral has better performance on classification accuracy than those classical algorithms. Deep fuzzy integral is the vertical generalization of classical fuzzy integral, which solves the crossover of data and promotes the classification performance.