Table 1 F-score and AUC for our four experiments using different hidden layer sizes
From: Combining lexical and context features for automatic ontology extension
Classification | Hidden layer sizes | 10 | 50 | 100 | 200 | ||||
|---|---|---|---|---|---|---|---|---|---|
Number of classes | F-score | AUC | F-score | AUC | F-score | AUC | F-score | AUC | |
Diseases | 2 | 94.65% | 95.31% | 94.83% | 95.97% | 95.32% | 96.06% | 94.49% | 95.99% |
Infectious disease | 5 | 95.65% | 95.01% | 96.01% | 95.74% | 95.43% | 95.22% | 95.68% | 96.42% |
Anatomical disease | 13 | 69.18% | 77.22% | 70.15% | 80.24% | 70.20% | 76.98% | 72.00% | 85.11% |
Infectious + anatomical diseases | 17 | 71.07% | 84.75% | 73.13% | 84.03% | 72.61% | 84.98% | 72.67% | 83.66% |