Distance metric learning for augmenting the method of nearest neighbors for ordinal classification with absolute and relative information

The performance of a classifier is often limited by the amount of labeled data (absolute information) available. In order to overcome this limitation, the incorporation of side information into the classification process has become a popular research topic in the field of machine learning. In this work, we propose a new method for ordinal classification that combines absolute information and a specific type of side information: relative information. In particular, this method exploits both types of information to learn an appropriate distance metric and subsequently incorporates the learned distance metric into the classical method of.. nearest neighbors. The experimental results show that the proposed method attains a good performance in terms of some of the most popular (ordinal) classification performance measures.