Table 3 Comparison of modified models
From: Antimicrobial peptide identification using multi-scale convolutional network
Model | SENS (%) | SPEC (%) | ACC (%) | MCC | auROC (%) |
|---|---|---|---|---|---|
Replacing embedding layer | 89.61 | 93.26 | 91.43 | 0.8282 | 96.75 |
Replacing multi-scale convolutional network | 89.75 | 91.15 | 90.44 | 0.8091 | 96.08 |
Replacing pooling1 with LSTM | 89.75 | 93.25 | 91.5 | 0.8305 | 96.27 |
Without pooling2 | 91.15 | 92.56 | 91.85 | 0.8371 | 96.3 |
Additional FC layers | 90.31 | 93.68 | 91.99 | 0.8403 | 97.09 |
proposed model | 91.01 | 93.64 | 92.41 | 0.8486 | 97.23 |