Text Categorization on Elec (test)
5.4Error RateAdversarial + Virtual Adversarial
Evaluation Results
| Method | Links | |
|---|---|---|
| Adversarial + Virtual Adversarial2016.05 | 5.4 | |
| Adversarial + Virtual AdversarialBackbone=bidirectional LSTM2016.05 | 5.45 | |
| Virtual Adversarial2016.05 | 5.54 | |
| Virtual AdversarialBackbone=bidirectional LSTM2016.05 | 5.55 | |
| One-hot bi-LSTMpretrained embeddings=CNN and bidirectional LSTM2016.05 | 5.55 | |
| Adversarial2016.05 | 5.61 | |
| One-hot CNN2016.05 | 5.87 | |
| oh-2LSTMpUnlabeled data usage=2x100-dim LSTM tv-embed.2016.02 | 6.08 | |
| Baseline2016.05 | 6.24 | |
| One-hot CNNpretrained embeddings=CNN2016.05 | 6.27 | |
| oh-CNNUnlabeled data usage=1x200-dim CNN tv-embed.2016.02 | 6.57 | |
| wv-2LSTMpUnlabeled data usage=200-dim word2vec scaled2016.02 | 6.76 | |
| wv-LSTMUnlabeled data usage=Pre-training2016.02 | 6.84 | |
| wv-2LSTMpUnlabeled data usage=300-dim Google News word2vec2016.02 | 7.64 | |
| NBLMdescription=Naive Bayes logisitic regression model2016.05 | 8.11 | |
| Transductive SVM2016.05 | 16.41 |