Universal Classification for Hidden
Neri Merhav · 1991
Binary hypotheses testing using empirically ob- served statistics is studied in the Neyman-Pearson formulation for the hidden Markov model (HMM). An asymptotically opti- mal decision rule is proposed and compared to the generalized likelihood ratio test (GLRT), which has been shown in earlier studies to be asymptotically optimal for simpler parametric families. The result can be applied to several types of HMM commonly used in speech recognition and communication appli- cations. Index Terms-Hypothesis testing, universal classification, hidden Markov model, large deviations, exponential families.