Convolutional Neural Network Committees for Handwritten Character Classification

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 S...

Full description

Saved in:
Bibliographic Details
Published in2011 International Conference on Document Analysis and Recognition pp. 1135 - 1139
Main Authors Ciresan, D. C., Meier, U., Gambardella, L. M., Schmidhuber, J.
Format Conference Proceeding
LanguageEnglish
Published IEEE 01.09.2011
Subjects
Character recognition
Committee
Computer architecture
Convolutional Neural Networks
Error analysis
Graphics Processing Unit
Handwriting recognition
Handwritten Character Recognition
Neural networks
NIST
Training
Online AccessGet full text
ISBN1457713500
9781457713507
ISSN1520-5363
DOI10.1109/ICDAR.2011.229

Cover

More Information
Summary: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.
ISBN:1457713500
9781457713507
ISSN:1520-5363
DOI:10.1109/ICDAR.2011.229