Learning-data selection mechanism through neural networks ensemble

In this paper we propose a model of neural networks ensemble consisting of a number of MLPs, that deals with an imperfect learning supervisor that occasionally produces incorrect teacher signals. It is known that a conventional unitary neural network will not learn optimally from this kind of supervisor. We consider that the imperfect supervisor generates two kinds of input-output relations, the correct relation and the incorrect one. The learning characteristics of the proposed model allows the ensemble to automatically train one of its members to learn only from the correct input-output relation, producing a neural network that can to some extent tolerate the imperfection of the super- visor.