Name
Abstract | ||
|---|---|---|
Neural networks are a powerful technology forclassification of visual inputs arising from documents.However, there is a confusing plethora of different neuralnetwork methods that are used in the literature and inindustry. This paper describes a set of concrete bestpractices that document analysis researchers can use toget good results with neural networks. The mostimportant practice is getting a training set as large aspossible: we expand the training set by adding a newform of distorted data. The next most important practiceis that convolutional neural networks are better suited forvisual document tasks than fully connected networks. Wepropose that a simple "do-it-yourself" implementation ofconvolution with a flexible architecture is suitable formany visual document problems. This simpleconvolutional neural network does not require complexmethods, such as momentum, weight decay, structure-dependentlearning rates, averaging layers, tangent prop,or even finely-tuning the architecture. The end result is avery simple yet general architecture which can yieldstate-of-the-art performance for document analysis. Weillustrate our claims on the MNIST set of English digitimages. |
Year | DOI | Venue |
|---|---|---|
2003 | 10.1109/ICDAR.2003.1227801 | ICDAR-1 |
Keywords | Field | DocType |
forvisual document task,neural network,visual document analysis,document analysis researcher,convolutional neural network,flexible architecture,convolutional neural networks applied,avery simple,document analysis,general architecture,best practices,suitable formany visual document,simpleconvolutional neural network,best practice,information processing,concrete,text analysis,support vector machines,digital image,neural networks,convolution,handwriting recognition | Computer vision,Architecture,Information processing,MNIST database,Convolution,Computer science,Convolutional neural network,Support vector machine,Handwriting recognition,Artificial intelligence,Artificial neural network,Machine learning | Conference |
ISSN | ISBN | Citations |
1520-5363 | 0-7695-1960-1 | 550 |
PageRank | References | Authors |
84.14 | 4 | 5 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Patrice Y. Simard | 1 | 1112 | 155.00 |
| Dave Steinkraus | 2 | 550 | 84.14 |
| John Platt | 3 | 6611 | 1100.14 |
| PY Simard | 4 | 550 | 84.14 |
| JC Platt | 5 | 550 | 84.14 |