Fig. 4
From: A new algorithm to train hidden Markov models for biological sequences with partial labels
Comparison results 3 with synthetic data when initialized with fully connected transition matrix State number = 7 and training sample size is 1600. Training Methods: Ground truth model—black; cBW—green; cBW + model_selection—blue; Scheffer et al.—red. Decoding Methods: Viterbi—cross tick mark; posterior-Viterbi—circle tick mark. Generated by Matlab 2020a