Sentence examples for maximum entropy classifier from inspiring English sources
The second-stage algorithm uses a maximum entropy classifier.
Predicate-argument structures are used along with other linguistic features with a maximum entropy classifier.
The tools include a parser, a part-of-speech tagger, a named-entity recognizer and a maximum entropy classifier among others (see full list).
Particularly maximum entropy classifier can still get an average accuracy rate as 97.0% after the introduction of the emergency domain terms.
The experiments show that, the introduction of emergency domain words will increase the average accuracy of maximum entropy classifier and KNN classifier by 4%to5%5%.
With CHI as evaluation function to select text features, the addition of emergency domain words, Maximum Entropy classifier and KNN classifier, we conduct a series of experiments on emergency event texts classification.
For example, the wsj-0-18-left3words-distsim.tagger wsj-0-18-left3words-distsim.tagger wsj-0-18-left3words-distsim.tagger wsj-0-18-left3words-distsim.tagger wsj-0-18-left3words-distsim.taggerximodelntropy classisiers; both are trainedirectlyut the same amount of data; both are in Java).
Other popular choices for text classification tasks is the use of maximum entropy classifiers (relying on the principle of choosing the most uniform distribution satisfying the constraints given by the training data) or support vector machines.
Maximum Entropy classifiers, (similar to naïve Bayes classifiers), estimate the conditional probability of the class label given the text fragment, that is, p(c| d), where d is an input text, and c denotes a class.
Table 4 shows the results obtained when applying Maximum Entropy classifiers to the dataset Frag_ FE the fragments on which all three annotators agreed on both Focus and Evidence.
These experiments were performed on each of the datasets Frag_ F, Frag _E, Frag_ C, Frag_ P and Frag_ T. We also show results of classifying the smaller dataset on which all annotators agreed on both the Focus and the Evidence (Frag_ FE), using the Maximum Entropy classifiers.
