Machine learning and statistics shape a novel path in archaeal promoter annotation

BMC Bioinformatics

Table 4 Results of the ANN-based classification

		Accuracy (%)	Precision (%)	Recall (%)	Specificity (%)
H. volcanii	Blocks	92.48	93.05	92.03	92.96
	Downstream	91.08	90.67	91.45	90.77
	Shuffled	84.55	84.86	84.27	84.94
S. solfataricus	Blocks	89.03	91.43	87.01	91.18
	Downstream	87.36	86.93	88.23	86.48
	Shuffled	86.63	84.56	88.27	85.17
T. kodakarensis	Blocks	94.96	93.39	96.21	93.83
	Downstream	91.35	91.69	91.31	91.46
	Shuffled	86.46	84.1	89.12	84.36

Each cell of this table contains the performance achieved by the best epoch for weight updating across the training dataset, i. e. the epochs were no longer increased when the convergence error became stable. For more details on the ANN simulation, see "Classification through an artificial neural network approach" and "Artificial Neural Network conveys a sturdier classification" sections

ISSN: 1471-2105