A method for training a feed-forward neural net model while targeting reduced nonlinearity | IEEE Conference Publication | IEEE Xplore