Maximum entropy discrimination (MED) feature subset selection for speech recognition

In this paper, we investigate the application of maximum entropy discrimination (MED) feature selection in speech recognition problems. We compare the MED algorithm with a classical wrapper feature selection algorithm and we propose a hybrid wrapper/MED algorithm. We experiment with the three approaches on a phoneme recognition task on the TIMIT database. Results show that the MED algorithm achieves error rates comparable with the wrapper algorithm, requiring a reduced computational charge. Furthermore, the use of a probabilistic framework shows that the MED algorithm gives very good results even with a very limited amount of data.