Learning cost-sensitive active classifiers

Most classification algorithms are “passive”, in that they assign a class label to each instance based only on the description given, even if that description is incomplete. By contrast, an active classifier can — at some cost — obtain the values of some unspecified attributes, before deciding upon a class label. This can be useful, for instance, when deciding whether to gather information relevant to a medical procedure or experiment. The expected utility of using an active classifier depends on both the cost required to obtain the values of additional attributes and the penalty incurred if the classifier outputs the wrong classification. This paper analyzes the problem of learning optimal active classifiers, using a variant of the probably-approximately-correct (PAC) model. After defining the framework, we show that this task can be achieved efficiently when the active classifier is allowed to perform only (at most) a constant number of tests. We then show that, in more general environments, this task of learning optimal active classifiers is often intractable.