Previous work has considered methods for learning projections of high-dimensional acoustic representations to lower dimensional spaces. In this paper we apply the neighborhood components analysis (NCA)  method to acoustic modeling in a speech recognizer. NCA learns a projection of acoustic vectors that optimizes a criterion that is closely related to the classification accuracy of a nearest-neighbor classifier. We introduce regularization into this method, giving further improvements in performance. We describe experiments on a lecture transcription task, comparing projections learned using NCA and HLDA . Regularized NCA gives a 0.7% absolute reduction in WER over HLDA, which corresponds to a relative reduction of 1.9%.
Bibliographic reference. Singh-Miller, Natasha / Collins, Michael / Hazen, Timothy J. (2007): "Dimensionality reduction for speech recognition using neighborhood components analysis", In INTERSPEECH-2007, 1158-1161.