Table 4 Assessing the efficiency of various optimizers and activation functions.
From: Neural network models for diagnosing recurrent aphthous ulcerations from clinical oral images
Model | Optimizer | Activation | Val_Accuracy | Precision | Recall | F1-score |
|---|---|---|---|---|---|---|
UlcerNet-1 | Adam with softmax | ReLU | 91 | 92 | 91 | 91 |
ELU | 90 | 91 | 91 | 91 | ||
GELU | 86 | 87 | 86 | 86 | ||
SELU | 93 | 94 | 93 | 93 | ||
UlcerNet-2 | RMSprop with softmax | ReLU | 89 | 90 | 89 | 89 |
ELU | 96 | 96 | 96 | 96 | ||
GELU | 93 | 93 | 93 | 93 | ||
SELU | 96 | 97 | 96 | 96 | ||
UlcerNet-3 | Adamax with softmax | ReLU | 94 | 95 | 95 | 95 |
ELU | 95 | 96 | 95 | 95 | ||
GELU | 92 | 93 | 93 | 91 | ||
SELU | 94 | 94 | 94 | 94 |
