A novel hierarchical Bayesian HMM for multi-dimensional discrete data

Authors:
Shigeru Motoi;Yohei Nakada;Toshie Misu;Takashi Matsumoto;Nobuyuki Yagi
Affiliations:
Waseda University, Tokyo, Japan;Waseda University, Tokyo, Japan;NHK (Japan Broadcasting Corporation), Tokyo, Japan;Waseda University, Tokyo, Japan;NHK (Japan Broadcasting Corporation), Tokyo, Japan
Venue:
AIA '08 Proceedings of the 26th IASTED International Conference on Artificial Intelligence and Applications
Year:
2008

Citing 7
Cited 0

A practical Bayesian framework for backpropagation networks

Neural Computation
Bayesian Learning for Neural Networks

Bayesian Learning for Neural Networks
Robust Tracking of Soccer Players Based on Data Fusion

ICPR '02 Proceedings of the 16 th International Conference on Pattern Recognition (ICPR'02) Volume 1 - Volume 1
Predictive automatic relevance determination by expectation propagation

ICML '04 Proceedings of the twenty-first international conference on Machine learning
Bayesian reconstructions and predictions of nonlinear dynamical systems via the hybrid Monte Carlo scheme

Signal Processing - Content-based image and video retrieval
Feature Selection Methods for Hidden Markov Model-Based Speech Recognition

ICPR '96 Proceedings of the 13th International Conference on Pattern Recognition - Volume 2
Reconstructions and predictions of nonlinear dynamical systems: ahierarchical Bayesian approach

IEEE Transactions on Signal Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper proposes a novel Bayesian Hidden Markov Model for multi-dimensional discrete time-series data. The proposed model has hyperparameters, which correspond to the dependencies of the data components on the hidden states. By adjusting these hyperparameters, the proposed model enables a reduction in negative influences from ineffective data components. This paper also describes an implementation method for the proposed model using the Markov Chain Monte Carlo method. The performance of the proposed model is evaluated via two examples.