Hierarchical error-correcting output codes based on SVDD

摘要

Error-correcting output codes (ECOC) can effectively reduce the multiclass to the binary and is attracting close attention, in which the construction of coding matrix based on data is the key to use ECOC to solve multiclass problems. An approach to the hierarchical error-correcting output codes based on support vector data description is presented in this paper. The main idea of the work is to construct the data-driven coding matrix with the help of support vector data description and binary tree. The support vector data description is used to measure the class separability quantitatively to obtain the inter-class separability matrix. And, a binary tree is built based on the matrixes from bottom to top. Then, each node of each layer is encoded to get the final hierarchical error-correcting output code. The independence of base classifiers trained by different encoding methods is compared in experiments. The results show that the proposed technique can promote the diversity of the base classifiers and enhance the classification accuracy.