Apparatus and method for improved classifier training
A non-transitory computer readable storage medium includes instructions to maintain an original training set of labeled documents, where the labeled documents correspond to a variety of topics. A new labeled document corresponding to a new topic is received. The original training set of labeled documents is modulated such that the new labeled document is over-represented with respect to the original training set. This results in a modulated training set. A classifier is trained with the modulated training set to form a trained classifier.
1. A non-transitory computer readable storage medium, comprising instructions to:
maintain an original training set of labeled documents, wherein the labeled documents correspond to a plurality of topics;
receive a new labeled document corresponding to a new topic;
modulate the original training set of labeled documents such that the new labeled document is over-represented with respect to the original training set to form a modulated training set, such that older topics with more examples are not over-represented with respect to the new labeled document; and
train a classifier with the modulated training set to form a trained classifier.
2. The non-transitory computer readable storage medium of claim 1 wherein the instructions to modulate include instructions to require a minimum number of labeled documents corresponding to the new topic.
3. The non-transitory computer readable storage medium of claim 1 further comprising executable instructions to apply documents to the trained classifier.
4. The non-transitory computer readable storage medium of claim 1 wherein the instructions to modulate include instructions to form a labeled training window with a size corresponding to the number of topics times the smallest number of labeled examples for one topic.
5. The non-transitory computer readable storage medium of claim 4 further comprising instructions to increase the labeled training window size until a minimum number of labeled examples occur in the labeled training window size.
6. The non-transitory computer readable storage medium of claim 4 further comprising instructions to increase the labeled training window size such that for any topic that has less than a minimum number of labeled examples, additional examples for that topic are added as if the training window were increased just for that topic.