We consider the problem of using nearest neighbor methods to provide a conditional probability estimate, P(y|a), when the number of labels y is large and the labels share some underlying structure. We propose a method for learning error-correcting output codes (ECOCs) to model the similarity between labels within a nearest neighbor framework. The learned ECOCs and nearest neighbor information are used to provide conditional probability estimates. We apply these estimates to the problem of acoustic modeling for speech recognition. We demonstrate an absolute reduction in word error rate (WER) of 0.9% (a 2.5% relative reduction in WER) on a lecture recognition task over a state-of-the-art baseline GMM model.
Natasha Singh-Miller (Massachusetts Institute of Technology)
Michael Collins (Columbia University)
Michael Collins is the Vikram S. Pandit Professor of computer science at Columbia University. His research is focused on topics including statistical parsing, structured prediction problems in machine learning, and applications including machine translation, dialog systems, and speech recognition. His awards include a Sloan fellowship, an NSF career award, and best paper awards at EMNLP (2002, 2004, and 2010), UAI (2004 and 2005), and CoNLL 2008.
More from the Same Authors
2012 Poster: Tensor Decomposition for Fast Parsing with Latent-Variable PCFGs »
Shay Cohen · Michael Collins
2011 Session: Oral Session 2 »
2011 Tutorial: Lagrangian Relaxation Algorithms for Inference in Natural Language Processing »
Alexander Rush · Michael Collins