i5mC-DCGA: an improved hybrid network framework based on the CBAM attention mechanism for identifying promoter 5mC sites

Table 1 Comparison of different network architecture models

Model frameworks	SN	SP	Acc	MCC	AUC
AdaBoost	0.6192	0.9655	0.9250	0.6183	0.7923
Random Forest	0.7537	0.9677	0.9427	0.7220	0.8607
ResNet	0.9702	0.9151	0.9215	0.7274	0.9770
Pri DenseNet	0.9754	0.9148	0.9219	0.7301	0.9774
Imp DenseNet	0.9843	0.9276	0.9342	0.7643	0.9795
Imp DenseNet + BiGRU	0.9719	0.9634	0.9644	0.8513	0.9849
GRU + Self-Attention	0.9836	0.9560	0.9592	0.8370	0.9806
Imp DenseNet + Self-Attention	0.9874	0.8976	0.9081	0.7035	0.9768
Imp DenseNet + BiGRU + Self-Attention	0.9806	0.9603	0.9626	0.8475	0.9837
Imp DenseNet + BiGRU + Self-Attention	0.9702	0.9652	0.9658	0.8558	0.9866

ISSN: 1471-2164