We show that the standard beam-search algorithm can be used as an efficient decoder for the global linear model of Zhang and Clark (2008) for joint word segmentation and POS-taggi...
Prediction of individual sequences is investigated for cases in which the decision maker observes a delayed version of the sequence, or is forced to issue his/her predictions a nu...
Unsupervised grammar induction is one of the most difficult works of language processing. Its goal is to extract a grammar representing the language structure using texts without a...
: This paper proposes twin prototype support vector machine (TVM), a constant space and sublinear time support vector machine (SVM) algorithm for online learning. TVM achieves its ...
We have implemented an incremental lexical acquisition mechanism that learns the meanings of previously unknown words from the context in which they appear, as a part of the proce...