OPTIMIZING MULTI-CLASS MULTIMEDIA DATA CLASSIFICATION USING NEGATIVE DATA

摘要

Techniques for optimizing multi-class image classification by leveraging negative multimedia data items to train and update classifiers are described. The techniques describe accessing positive multimedia data items of a plurality of multimedia data items, extracting features from the positive multimedia data items, and training classifiers based at least in part on the features. The classifiers may include a plurality of model vectors each corresponding to one of the individual labels. The system may iteratively test the classifiers using positive multimedia data and negative multimedia data and may update one or more model vectors associated with the classifiers differently, depending on whether multimedia data items are positive or negative. Techniques for applying the classifiers to determine whether a new multimedia data item is associated with a topic based at least in part on comparing similarity values with corresponding statistics derived from classifier training are also described.

主权项

1. A system comprising:
computer-readable media; one or more processors; and one or more modules stored in the computer-readable media and executable by the one or more processors to perform operations comprising:
accessing a multimedia data item;extracting features from the multimedia data item;applying a classifier to the features to determine similarity values corresponding to individual labels of a plurality of labels, the classifier including a plurality of model vectors each corresponding to one of the individual labels;determining whether the multimedia data item is a positive multimedia data item or a negative multimedia data item; andupdating at least one model vector of the plurality of model vectors, wherein updating the at least one model vector comprises applying a first update for the positive multimedia data item and a second update for the negative multimedia data item.