Deep, Big, Simple Neural Nets for Handwritten Digit Recognition
IDSIA, 6928 Manno-Lugano, Switzerland
Neural Computation, Vol. 22, No. 12. (21 September 2010), pp. 3207-3220
@article{ciresan2010deep,
title={Deep, Big, Simple Neural Nets for Handwritten Digit Recognition.},
author={Cire{c{s}}an, DC and Meier, U. and Gambardella, LM and Schmidhuber, J.},
journal={Neural computation},
issn={1530-888X},
year={2010}
}
Good old online backpropagation for plain multilayer perceptrons yields a very low 0.35% error rate on the MNIST handwritten digits benchmark. All we need to achieve this best result so far are many hidden layers, many neurons per layer, numerous deformed training images to avoid overfitting, and graphics cards to greatly speed up learning. Good old online backpropagation for plain multilayer perceptrons yields a very low 0.35% error rate on the MNIST handwritten digits benchmark. All we need to achieve this best result so far are many hidden layers, many neurons per layer, numerous deformed training images to avoid overfitting, and graphics cards to greatly speed up learning.
February 9, 2011 by hgpu
