Inductive Confidence Machines for Regression

摘要: The existing methods of predicting with confidence give good accuracy and values, but quite often are computationally inefficient. Some partial solutions have been suggested in the past. Both original method these were based on transductive inference. In this paper we make a radical step replacing inference inductive define what call Inductive Confidence Machine (ICM); our main concern is use ICM regression problems. algorithm proposed Ridge Regression procedure (which usually used for outputting bare predictions) much faster than techniques. approach described may be only option available when dealing large data sets.