A Coupled k-Nearest Neighbor Algorithm for Multi-label Classification

Chunming Liu, Longbing Cao

Abstract: ML- $$k$$ NN is a well-known algorithm for multi-label classification. Although effective in some cases, ML- $$k$$ NN has some defect due to the fact that it is a binary relevance classifier which only considers one label every time. In this paper, we present a new method for multi-label classification, which is based on lazy learning approaches to classify an unseen instance on the basis of its $$k$$ nearest neighbors. By introducing the coupled similarity between class labels, the proposed method exploits the correlations between class labels, which overcomes the shortcoming of ML- $$k$$ NN. Experiments on benchmark data sets show that our proposed Coupled Multi-Label $$k$$ Nearest Neighbor algorithm (CML- $$k$$ NN) achieves superior performance than some existing multi-label classification algorithms.