Maximum likelihood multiple subspace projections for hidden Markov models.

Maximum likelihood multiple subspace projections for hidden Markov models.

The first stage in many pattern recognition tasks is to generate a good set of features from the observed data. Usually, only a single feature space is used. However, in some complex pattern recognition tasks the choice of a good feature space may vary depending on the signal content. An example is...

Ausführliche Beschreibung

Bibliographische Detailangaben
Veröffentlicht in:IEEE Transactions on speech and audio processing 10, 2 (2002).
1. Verfasser: Gales, M.J.F
Format: Artikel
Sprache:English
Schlagworte:
Computational cost.
Estimation formulae.
Feature space.
Hidden Markov models.
Likelihood consistency.
Maximum likelihood framework.
Maximum likelihood multiple subspace projections.
Multiple subspaces.
Pattern recognition tasks.
Phone dependent feature subspaces.
Projection schemes.
Signal content.
Speech recognition.