Research Article
Deep Learning Structure for Cross-Domain Sentiment Classification Based on Improved Cross Entropy and Weight
Table 9
Results of different dropout values.
| ā | Dropout | Accuracy (%) | Loss (%) | Time (s) |
| LSTM-BO | 0.1 | 81.62 | 0.2348 | 1265 | 0.2 | 82.14 | 0.2267 | 1136 | 0.3 | 82.06 | 0.2285 | 1148 | 0.4 | 82.05 | 0.2274 | 1143 | 0.5 | 81.98 | 0.2354 | 1254 |
| CNN-BO | 0.1 | 87.72 | 0.1833 | 1249 | 0.2 | 88.74 | 0.1637 | 1172 | 0.3 | 88.72 | 0.1590 | 1184 | 0.4 | 88.39 | 0.1684 | 1227 | 0.5 | 87.43 | 0.1842 | 1265 |
|
|