PropBank has been widely used as training data for Semantic Role Labeling. However, because this training data is taken from the WSJ, the resulting machine learning models tend to...
Feature selection, as a preprocessing step to machine learning, has been very effective in reducing dimensionality, removing irrelevant data, increasing learning accuracy, and imp...
Mining bilingual data (including bilingual sentences and terms1 ) from the Web can benefit many NLP applications, such as machine translation and cross language information retrie...
Long Jiang, Shiquan Yang, Ming Zhou, Xiaohua Liu, ...
The asymptotic behavior of stochastic gradient algorithms is studied. Relying on some results of differential geometry (Lojasiewicz gradient inequality), the almost sure pointconve...
With the development of telecom business, customer churn prediction becomes more and more important. An outstanding issue in customer churn prediction is high dimensional problem....