Ninth International Conference on Spoken Language Processing

Pittsburgh, PA, USA
September 17-21, 2006

On a Greedy Learning Algorithm for dPLRM with Applications to Phonetic Feature Detection

Tor André Myrvoll (1), Tomoko Matsui (2)

(1) NTNU, Norway; (2) Institute of Statistical Mathematics, Japan

In this work we investigate the use of a greedy training algorithm for the dual Penalized Logistic Regression Machine (dPLRM), and our target application is detection of broad class phonetic features. The use of a greedy training algorithm is meant to alleviate the infeasible memory and computational demands that arises during the learning phase when the amount of training data increases. We show that using only a subset of the training data, chosen in a greedy manner, we can achieve as good as or better performance as when using the full training set. We can also train dPLRMs using data sets that are significantly larger than what our current computational resources can accommodate when using non-greedy approaches.

Full Paper

Bibliographic reference.  Myrvoll, Tor André / Matsui, Tomoko (2006): "On a greedy learning algorithm for dPLRM with applications to phonetic feature detection", In INTERSPEECH-2006, paper 2063-Wed1FoP.10.