PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Sequence Classification with Input-Output Hidden Markov Models
Silvia Chiappa and Samy Bengio
(2004) Technical Report. IDIAP Research Institute, Martigny, Switzerland.


We present a training and testing method for Input-Output Hidden Markov Model that is particularly suited for classification of sequences in which class information accumulates over time. We discuss two such cases: the discrimination of mental tasks from sequences of EEG features, common in Brain Computer Interface research, and phoneme classification from sequences of acoustic features for speech recognition. The objective function is modified so that training focuses on the improvement of classification accuracy. For both tasks the algorithm performs significantly better than the alternative solution proposed in the literature, specifically designed for other types of sequences.

EPrint Type:Monograph (Technical Report)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
Brain Computer Interfaces
ID Code:674
Deposited By:Silvia Chiappa
Deposited On:29 December 2004