Table 5 Improvements of cBW + model selection, cBW alone vs Scheffer et al. in cases of correct initial transition matrix for synthetic data with posterior-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 | 7.46/8.01% | 2.8E−02/1.7E−02 | 7.64/8.11% |
3/2000 | 8.09/8.52% | 3.0E−02/1.6E−02 | 8.25/8.60% |
3/2400 | 8.32/8.67% | 3.7E−02/5.9E−03 | 8.49/8.73% |
3/2800 | 8.58/8.99% | 4.9E−02/1.2E−03 | 8.79/9.09% |
5/1600 | 9.52/10.62% | 1.1E−01/5.1E−02 | 9.45/10.59% |
5/2000 | 10.53/11.55% | 6.0E−02/8.2E−03 | 10.47/11.63% |
5/2400 | 11.51/12.58% | 2.0E−02/3.5E−04 | 11.44/12.58% |
5/2800 | 12.49/13.55% | 1.7E−02/8.0E−06 | 12.55/13.61% |
7/1600 | 8.75/10.19% | 2.9E−02/2.6E−02 | 8.27/9.77% |
7/2000 | 10.50/11.99% | 2.1E−02/4.2E−03 | 9.82/11.31% |
7/2400 | 11.15/12.47% | 6.9E−02/7.8E−04 | 10.69/12.11% |
7/2800 | 12.36/13.75% | 5.2E−02/1.1E−05 | 12.05/13.57% |