Many machine learning methods have recently been applied to natural language processing tasks. Among them, the Winnow algorithm has been argued to be particularly suitable for NLP...
Stochastic gradient descent (SGD) uses approximate gradients estimated from subsets of the training data and updates the parameters in an online fashion. This learning framework i...