Dynamic tunneling based regularization in feedforward neural networks

This paper presents a new regularization method based on dynamic tunneling for enhancing generalization capability of multilayered neural networks. Th e proposed method enables escape through undesired sub-optimal solutions on the composite error surface by means of dynamic tunneling. Undesired sub-o ptimal solutions may be increased or introduced from regularized objective function. Hence, the proposed method is capable of enhancing the regulariza tion property without getting stuck at sub-optimal values in search space. The regularization property and escape from the sub-optimal values have bee n demonstrated through computer simulations on two examples. (C) 2001 Elsev ier Science B.V. All rights reserved.