Discriminant projections embedding for nearest neighbor classification

Abstract

In this paper we introduce a new embedding technique to linearly project labeled data samples into a new space where the performance of a Nearest Neighbor classifier is improved. The approach is based on considering a large set of simple discriminant projections and finding the subset with higher classification performance. In order to implement the feature selection process we propose the use of the adaboost algorithm. The performance of this technique is tested in a multiclass classification problem related to the production of cork stoppers for wine bottles. © Springer-Verlag 2004.