Classification Among Hidden Markov Models

On the Sequential Probability Ratio Test in Hidden Markov Models

“…An HMM is typically viewed as a producer of a finite or infinite word of emitted observations. For example, starting in q 1 , the probability of producing a word with prefix aba is 1 3 • 2 3 • 2 3 , whereas starting in q 2 , the probability of aba is 2 3…”

“…Call distributions π 1 , π 2 distinguishable if d(π 1 , π 2 ) = 1. Distinguishability was used for runtime monitoring [23] and diagnosability [4,2] of stochastic systems.…”