... dogs without ever seeing all in the training data We expect the second training regime to be more difficult, not just because there is less training data, but also because the trained classifier is ... below) In balAPincAN |= N , we use the AN |= N data set as training data and pick the t that maximizes F on this training set We use the balAPinc measure as a reference point because, on the evidence ... other 29 pairs in Table as training data Thus, for example, the classifier must discover all dogs |= some dogs without seeing any all N |= some N instance in the training data In the second (SVMquantifier-out...