针对多类分类问题,利用纠错输出编码作为分解框架,把多类问题转化为多个二类问题加以解决;同时提出一种基于证据理论的解码策略,把每一个二分器的输出作为证据之一进行融合,并讨论在两种编码类型(二元和三元编码矩阵)下证据融合的不同策略.通过实验分别对UCI数据集和3种一维距离像数据集进行测试,并与几种经典的解码方法进行比较,验证了所提出的方法能有效提高纠错输出编码特别是三元编码矩阵的分类正确率. In order to solve the problem of multi-class classification, the error-correcting output coding is used as the decomposition framework to solve the problem of converting multiple classes of problems into multiple class II problems. At the same time, a decoding strategy based on evidence theory is proposed, with the output of each classifier as evidence We discuss the different strategies of evidence fusion in the two coding types (binary and ternary coding matrices) .The UCI dataset and three kinds of one-dimensional distance image datasets are tested by experiments and compared with several The comparison of the classical decoding methods shows that the proposed method can effectively improve the classification accuracy of the error-correcting output codes, especially the ternary coding matrix.