Figure 2
From: New decoding algorithms for Hidden Markov Models using distance measures on labellings
Computing the probability of a footprint. To compute the probability of a footprint f = f1, ..., f k , we create a group of states G i for each entry in f, corresponding to the states in M labelled f i . A path compatible with the footprint must first go from x0 to one of the states in G1, then eventually go to G2, and on to G k . Transitions are only allowed from G i to G i or Gi+1.