Evaluation of distance-based techniques for data projection and dimensionality reduction

Today there are a few distance-based methods that cope with the problem of projecting high dimensional data to low space. These methods are more or less described on their own. In this paper, we explain the fundamentals of distance-based data projection and further derive a general framework. We describe today's methods by pointing out their common aspects and differences. We investigate their strengths and weaknesses and carry out experiment to applying these methods to recover a 2D synthetic data set wrapped in 3D in several ways. Features of these methods are discussed and summarized through the experiment.