Decision-tree based feature-space quantization for fast gaussian computation

Abstract

The feature space of a speech recognizer is quantized using a binary decision-tree. Each node of the decision tree represents a region of the feature space, and is characterized by a hyperplane that subdivides the region corresponding to the current node into two non-overlapping regions corresponding to the two children of the current node. Given a test feature vector, the process of finding the region that it lies in involves traversing this binary decision tree. Results of the experiments that show that the gaussian computation time can be reduced by as much as a factor of 20 with negligible degradation in accuracy are presented.