Transduction with Confidence and Credibility

In this paper we follow the same general ideology as in [Gammerman et al., 1998], and describe a new transductive learning algorithm using Support Vector Machines. The algorithm presented provides confidence values for its predicted classifications of new examples. We also obtain a measure of “credibility” which serves as an indicator of the reliability of the data upon which we make our prediction. Experiments compare the new algorithm to a standard Support Vector Machine and other transductive methods which use Support Vector Machines, such as Vapnik’s margin transduction. Empirical results show that the new algorithm not only produces confidence and credibility measures, but is comparable to, and sometimes exceeds the performance of the other algorithms.