Title :
Convolutional Neural Network Committees for Handwritten Character Classification
Author :
Dan Claudiu Ciresan;Ueli Meier;Luca Maria Gambardella;Jurgen Schmidhuber
Author_Institution :
IDSIA, USI, Manno-Lugano, Switzerland
Abstract :
In 2010, after many years of stagnation, the MNIST handwriting recognition benchmark record dropped from 0.40% error rate to 0.35%. Here we report 0.27% for a committee of seven deep CNNs trained on graphics cards, narrowing the gap to human performance. We also apply the same architecture to NIST SD 19, a more challenging dataset including lower and upper case letters. A committee of seven CNNs obtains the best results published so far for both NIST digits and NIST letters. The robustness of our method is verified by analyzing 78125 different 7-net committees.
Keywords :
"NIST","Training","Error analysis","Handwriting recognition","Neural networks","Character recognition","Computer architecture"
Conference_Titel :
Document Analysis and Recognition (ICDAR), 2011 International Conference on
Print_ISBN :
978-1-4577-1350-7
DOI :
10.1109/ICDAR.2011.229