Skip to main content

Table 3 Improvements of cBW + model selection, cBW alone vs Scheffer et al. in cases of correct initial transition matrix for synthetic data with Viterbi algorithm

From: A new algorithm to train hidden Markov models for biological sequences with partial labels

State #/training sample

Average improvements of cBW + model selection in setting 1/2

Average p value of cBW + model selection in setting 1/2

Average improvements of cBW alone in setting 1/2

3/1600

8.07/8.57%

2.3E−04/4.7E−07

8.11/8.56%

3/2000

8.58/9.05%

1.9E−06/6.5E−09

8.63/9.03%

3/2400

8.93/9.24%

1.6E−07/1.0E−09

8.97/9.20%

3/2800

8.87/9.31%

1.3E−08/1.5E−09

8.94/9.26%

5/1600

11.99/13.24%

1.7E−02/4.5E−06

11.76/13.08%

5/2000

13.07/14.20%

4.1E−02/3.4E−06

12.87/14.11%

5/2400

13.22/14.59%

2.0E−02/1.2E−05

12.94/14.35%

5/2800

13.89/15.20%

4.1E−02/1.6E−07

13.85/15.16%

7/1600

7.85/9.37%

6.7E−02/3.5E−02

6.04/7.34%

7/2000

9.75/11.28%

5.6E−03/4.1E−03

7.93/9.32%

7/2400

10.50/12.10%

1.8E−02/1.9E−02

8.99/10.53%

7/2800

11.39/12.95%

1.4E−03/1.9E−04

9.75/11.29%