Deep big multilayer perceptrons for digit recognition

Dan Claudiu Cireşan, Ueli Meier, Luca Maria Gambardella, Jürgen Schmidhuber

Research output: Contribution to journalArticlepeer-review

33 Scopus citations

Abstract

The competitive MNIST handwritten digit recognition benchmark has a long history of broken records since 1998. The most recent advancement by others dates back 8 years (error rate 0.4 old on-line back-propagation for plain multi-layer perceptrons yields a very low 0.35% error rate on the MNIST handwritten digits benchmark with a single MLP and 0.31% with a committee of seven MLP. All we need to achieve this until 2011 best result are many hidden layers, many neurons per layer, numerous deformed training images to avoid overfitting, and graphics cards to greatly speed up learning. © Springer-Verlag Berlin Heidelberg 2012.
Original languageEnglish (US)
Pages (from-to)581-598
Number of pages18
JournalLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume7700 LECTURE NO
DOIs
StatePublished - Jan 1 2012
Externally publishedYes

Fingerprint

Dive into the research topics of 'Deep big multilayer perceptrons for digit recognition'. Together they form a unique fingerprint.

Cite this