More Information
Monthly
288 pp. per issue
6 x 9, illustrated
ISSN
0899-7667
E-ISSN
1530-888X
2014 Impact factor:
2.21
List of All Issues » Issue TOC

Neural Computation

December 2010, Vol. 22, No. 12, Pages 3207-3220
(doi: 10.1162/NECO_a_00052)
© 2010 Massachusetts Institute of Technology
Deep, Big, Simple Neural Nets for Handwritten Digit Recognition
Article PDF (555.86 KB)
Dan Claudiu Cireşan
IDSIA, 6928 Manno-Lugano, Switzerland, and University of Lugano, 6900 Lugano, Switzerland, and Sculoa universitaria professionale della svizzera italiana, 6928 Manno-Lugano, Switzerland dan@idsia.ch
Ueli Meier
IDSIA, 6928 Manno-Lugano, Switzerland, and University of Lugano, 6900 Lugano, Switzerland, and Sculoa universitaria professionale della svizzera italiana, 6928 Manno-Lugano, Switzerland ueli@idsia.ch
Luca Maria Gambardella
IDSIA, 6928 Manno-Lugano, Switzerland, and University of Lugano, 6900 Lugano, Switzerland, and Sculoa universitaria professionale della svizzera italiana, 6928 Manno-Lugano, Switzerland luca@idsia.ch
Jürgen Schmidhuber
IDSIA, 6928 Manno-Lugano, Switzerland, and University of Lugano, 6900 Lugano, Switzerland, and Sculoa universitaria professionale della svizzera italiana, 6928 Manno-Lugano, Switzerland juergen@idsia.ch
Abstract

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.