Fig. 1
From: A new algorithm to train hidden Markov models for biological sequences with partial labels
Example of constrained Baum–Welch’s forward/backward dynamic programming table construction Position 1–2 is the case of labelled position to unlabelled position. Position 3–4 is the case of unlabelled position to labelled position. Position 4–5 is the case of labelled position to labelled position. Dashed lines indicate state transitions. Generated by Google Drawings