Robust estimation of discrete hidden Markov model parameters using the entropy-based feature-parameter weighting and source-quantization modeling

We propose a new variant of the discrete hidden Markov model (DHMM) in which the output distribution is estimated by state-dependent source quantizing modeling and the output probability is weighted by the entropy of each feature-parameter at a state. The state-dependent source is represented as a s...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Artificial intelligence in engineering Jg. 12; H. 3; S. 243 - 252
Hauptverfasser: Choi, Hwan Jin, Yun, Sung Jin, Oh, Yung Hwan
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Oxford Elsevier Ltd 01.07.1998
Elsevier
Schlagworte:
Applied sciences
Artificial intelligence
Computer science; control theory; systems
entropy
Exact sciences and technology
feature-parameter weighting
fuzzy objective function
hidden Markov model
Learning and adaptive systems
Speech and sound recognition and synthesis. Linguistics
state-dependent source quantization modeling
state-dependent source quantization modeling
fuzzy objective function
entropy
hidden Markov model
feature-parameter weighting
Parameter estimation
Expert system
Quantization
Entropy
Discrete system
Markov model
Pattern recognition
Experimental study
Modeling
Fuzzy logic
System performance
Speech recognition
Objective function
Dynamic programming
Artificial intelligence
Ambiguity
ISSN:0954-1810
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:We propose a new variant of the discrete hidden Markov model (DHMM) in which the output distribution is estimated by state-dependent source quantizing modeling and the output probability is weighted by the entropy of each feature-parameter at a state. The state-dependent source is represented as a state-dependent quantized vector which is regarded as a variant of a representative vector at a state and its own codeword distribution, and the output distribution is derived by these state-dependent sources which will exist at a state. In addition, entropy-based feature-parameter weighting is proposed to reflect the different importance of each feature-parameter in a state, and the fuzzy function is applied to transform an entropy value into a feature-parameter weighting factor. From experiments, we found that proposed methods have shown an improvement of 5.6%, which indicates the effectiveness of proposed models in the robust estimation of output probabilities for DHMMs.
Bibliographie:ObjectType-Article-2
SourceType-Scholarly Journals-1
ObjectType-Feature-1
content type line 23
ISSN:0954-1810
DOI:10.1016/S0954-1810(97)00026-5