The Classification of Noisy Sequences Generated by Similar HMMs
The method for classification performance improvement using hidden Markov models (HMM) is proposed. The k-nearest neighbors (kNN) classifier is used in the feature space produced by these HMM. Only the similar models with the noisy original sequences assumption are discussed. The research results on simulated data for two-class classification problem are presented.
KeywordsHidden Markov Model Derivation k Nearest Neighbors
- 2.Cappé, O.: Ten years of HMMs. CNRS, LTCI & ENST, Dpt. TSI, http://perso.telecom-paristech.fr/~cappe/docs/hmmbib.html
- 3.Mottl, V.V., Muchnik, I.B.: Hidden Markov Models in Structural Signal Analysis Moscow, Russia (1999) (in Russian)Google Scholar
- 4.Zagorujko, N.G.: Applied methods of analysis of data and knowledge. Novosibirsk, Russia (1999) (in Russian)Google Scholar
- 5.Chen, L., Man, H.: Combination of Fisher Scores and Appearance Based by Features For Face. In: Proc. of the 2003 ACM SIGMM Workshop on Biometrics Methods and Applications, Berkeley, California, USA, pp. 74–81 (2003)Google Scholar