An Extension of Multi-label Binary Relevance Models Based on Randomized Reference Classifier and Local Fuzzy Confusion Matrix

Abstract

In this paper we addressed the issue of applying a stochastic classifier and a local, fuzzy confusion matrix under the framework of multi-label classification. We proposed a novel solution to the problem of correcting Binary Relevance ensembles. The main step of the correction procedure is to compute label-wise competence and cross-competence measures, which model error pattern of the underlying classifier. The method was evaluated using 20 benchmark datasets. In order to assess the efficiency of the introduced model, it was compared against 3 state-of-the-art approaches. The comparison was performed using 4 different evaluation measures. Although the introduced algorithm, as its base algorithm – Binary Relevance, is insensitive to dependencies between labels, the conducted experimental study reveals that the proposed algorithm outperform other methods in terms of Hamming-loss and False Discovery Rate.