Control of the error signals by self-awareness in committee machines

Abstract

It is certain that the individual learners should be different from each other in order for a committee machine to reach the better performance. However, differences alone among the individual learners are not enough for the committee machine to predict well on the unknown data. It would be essential for each individual learner to be able to decide whether to learn to be different or not to the other individuals on each given example. One way to implement such decision is through self-awareness. Self-awareness makes the individual learners in the committee machine be even more flexible during the learning process. With self-awareness, an individual learner could choose to go slower to the correct output by scaling down the error signals, or leave away faster from the correct output on a given data. In this paper, negative correlation learning with the scaled error signals were tested on the two medical data sets to show how important it is to adjust the error signals by the individual learners themselves in the committee machines.