INDUCTIVE POLICY - THE PRAGMATICS OF BIAS SELECTION

This paper extends the currently accepted model of inductive bias by i dentifying six categories of bias and separates inductive bias from th e policy for its selection (the inductive policy). We analyze existing ''bias selection'' systems, examining the similarities and difference s in their inductive policies, and identify three techniques useful fo r building inductive policies. We then present a framework for represe nting and automatically selecting a wide variety of biases and describ e experiments with an instantiation of the framework addressing variou s pragmatic tradeoffs of time, space, accuracy, and the cost of errors . The experiments show theta common framework can be used to implement policies for a variety of different types of bias selection, such as parameter selection, term selection, and example selection, using simi lar techniques. The experiments also show that different tradeoffs can be made by the implementation of different policies; for example, fro m the same data different rule sets can be learned based on different tradeoffs of accuracy versus the cost of erroneous predictions.