“Yang Wang 2016-03-28”版本间的差异
来自cslt Wiki
| 第1行： | 第1行： | ||
Weekly Report: | Weekly Report: | ||
* replaced maximum-entropy classifier with DNN and SVM. | * replaced maximum-entropy classifier with DNN and SVM. | ||
| − | found that SVM can improve performance significantly. | + | found that SVM can improve performance significantly. |
* add vectors that produced by a margin-based model as a feature. | * add vectors that produced by a margin-based model as a feature. | ||
* prepared the data for machine translation. | * prepared the data for machine translation. | ||
2016年3月28日 (一) 00:40的版本
Weekly Report:
- replaced maximum-entropy classifier with DNN and SVM.
found that SVM can improve performance significantly.
- add vectors that produced by a margin-based model as a feature.
- prepared the data for machine translation.
- started running the experiment.
To Do Next Week:
- further improve the performance of the algorithm in thread detection
- conduct experiment of machine translation
