The competitive MNIST handwritten digit recognition benchmark has a longhistory of broken records since 1998. The most recent substantial improvementby others dates back 7 years (error rate 0.4%) . Recently we were able tosignificantly improve this result, using graphics cards to greatly speed uptraining of simple but deep MLPs, which achieved 0.35%, outperforming all theprevious more complex methods. Here we report another substantial improvement:0.31% obtained using a committee of MLPs.
展开▼