... parameter has on the error rate, and then modifies the parameters to reduce the error rate based on this prediction Linear Models, the Perceptron Algorithm, and Conditional Random Fields This section ... framework, global linear models, and two parameter estimation methods within the framework, the perceptron algorithm and a method based on conditional random fields The linear models we describe ... then the trigram wi−2 wi−1 wi is a feature, as is the bigram wi−1 wi and the unigram wi In this case, the weight on the transition wi leaving state h must be the sum of the trigram, bigram and...