Margin trees for highdimensional classification

We propose a method for the classification of more than two classes, from high-dimensional features. Our approach is to build a binary decision tree in a top-down manner, using the optimal margin classifier at each split. We implement an exact greedy algorithm for this task, and compare its performance to less greedy procedures based on clustering of the matrix of pairwise margins. We compare the performance of the “margin tree ” to the closely related “all-pairs ” (one versus one) support vector machine, and nearest centroids on a number of cancer microarray data sets. We also develop a simple method for feature selection. We find that the margin tree has accuracy that is competitive with other methods and offers additional interpretability in its putative grouping of the classes