ON THE USE OF MLP-DISTANCE TO ESTIMATE POSTERIOR PROBABILITIES BY KNN FOR SPEECH RECOGNITION
Ana I. García Moral, Carmen Peláez Moreno, Hervé A. Bourlard · 2006
In this work we try to estimate a posteriori probabilities needed for speech recognition by the K-Nearest-Neighbors rule (KNN), using a Multi-Layered Perceptrom (MLP) to obtain the distances between neighbors. Thus, we can distinguish two different works: on the one hand, we es-timate a posteriori probabilities using KNN with the aim of using them in a speech recognition system [1][2]; and, on the other hand, we propose a new distance measure, MLP-distance. 1.