Softprop: softmax neural network backpropagation learning

@article{Rimer2004SoftpropSN,
  title={Softprop: softmax neural network backpropagation learning},
  author={M. Rimer and T. Martinez},
  journal={2004 IEEE International Joint Conference on Neural Networks (IEEE Cat. No.04CH37541)},
  year={2004},
  volume={2},
  pages={979-983 vol.2}
}
Multi-layer backpropagation, like many learning algorithms that can create complex decision surfaces, is prone to overfitting. Softprop is a novel learning approach presented here that is reminiscent of the softmax explore-exploit Q-learning search heuristic. It fits the problem while delaying settling into error minima to achieve better generalization and… CONTINUE READING