Improved method of handwritten digit recognition tested on MNIST database

Kussul, Ernst; Baidyk, Tatiana

doi:10.1016/j.imavis.2004.03.008

Cited by 177 publications

(77 citation statements)

References 8 publications

(22 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This implies that appropriateness of DOLs and allocation of their 0.5 rates tend to insert, for CNNs with a few ConvLs, just a single DOL before the last ConvL. This rule of rationally allocating DOLs excludes inserting a DOL after every single ConvL, unless the dataset is very simple (like the MNIST dataset [7], [8], [14], [17] resembling EEACL26), where overfitting is likelier. Figure 18 shows error rates for the EEACL26 dataset by using the rule versus using four other versions.…”

Section: The Rule Of Rationally Allocating Dolsmentioning

confidence: 99%

Appropriateness of Dropout Layers and Allocation of Their 0.5 Rates across Convolutional Neural Networks for CIFAR-10, EEACL26, and NORB Datasets

Romanuke

2017

Applied Computer Systems

View full text Add to dashboard Cite

-A technique of DropOut for preventing overfitting of convolutional neural networks for image classification is considered in the paper. The goal is to find a rule of rationally allocating DropOut layers of 0.5 rate to maximise performance. To achieve the goal, two common network architectures are used having either 4 or 5 convolutional layers. Benchmarking is fulfilled with CIFAR-10, EEACL26, and NORB datasets. Initially, series of all admissible versions for allocation of DropOut layers are generated. After the performance against the series is evaluated, normalized and averaged, the compromising rule is found. It consists in non-compactly inserting a few DropOut layers before the last convolutional layer. It is likely that the scheme with two or more DropOut layers fits networks of many convolutional layers for image classification problems with a plenty of features. Such a scheme shall also fit simple datasets prone to overfitting. In fact, the rule "prefers" a fewer number of DropOut layers. The exemplary gain of the rule application is roughly between 10 % and 50 %.

show abstract

Section: The Rule Of Rationally Allocating Dolsmentioning

confidence: 99%

Appropriateness of Dropout Layers and Allocation of Their 0.5 Rates across Convolutional Neural Networks for CIFAR-10, EEACL26, and NORB Datasets

Romanuke

2017

Applied Computer Systems

View full text Add to dashboard Cite

show abstract

“…Various methods have been proposed and high recognition rates are reported, for the recognition of English handwritten digits (Berkes, 2005;Liu et al, 2004;Kussul and Baidyk, 2004;Tang, 2006). In recent years, many researchers have addressed the recognition of Arabic text, including Arabic numerals (Al-Omari and AlJarrah, 2004;Bouslama, 1999;Salourn, 2001;Salah et al, 2002;Alma'adeed et al, 2004;Touj et al, 2005 (Mahmoud and Awaida, 2009).…”

Section: Discussionmentioning

confidence: 99%

Recognition of Hindi (Arabic) Handwritten Numerals

M.¹

2012

American Journal of Engineering and Applied Sciences

View full text Add to dashboard Cite

Recognition of handwritten numerals has been one of the most challenging topics in image processing. This is due to its contributions in the automation process in several applications. The aim of this study was to build a classifier that can easily recognize offline handwritten Arabic numerals to support those applications that are deal with Hindi (Arabic) numerals. A new algorithm for Hindi (Arabic) Numeral Recognition is proposed. The proposed algorithm was developed using MATLAB and tested with a large sample of handwritten numeral datasets for different writers in different ages. Pattern recognition techniques are used to identify Hindi (Arabic) handwritten numerals. After testing, high recognition rates were achieved, their ranges from 95% for some numerals and up to 99% for others. The proposed algorithm used a powerful set of features which proved to be effective in the recognition of Hindi (Arabic) numerals.

show abstract

“…Also, in [7], a technique utilizes a number of statistical methods to perform machine print recognition. In addition, several approaches that are based on Neural Networks and Support Vector Machine (SVM) have been investigated for recognition of on-line and off-line handwritten Arabic and Hindi numerals [8][9][10][11][12][13][14][15]. Likewise, Hidden Markov Models have also been adopted for recognition of off-line handwritten numerals [16].…”

Section: Literature Reviewmentioning

confidence: 99%

Off-Line Arabic (Indian) Numbers Recognition Using Expert System

Malallah¹,

Saeed²,

Aziz³

et al. 2016

ijacsa

View full text Add to dashboard Cite

Abstract-This paper proposes an effective approach to automatic recognition of printed Arabic numerals which are extracted from digital images. First, the input image is normalized and pre-processed to an acceptable form. From the preprocessed image, components of the words are segmented into individual objects representing different numbers. Second, the numerical recognition is performed using an expert system based on a set of if-else rules, where each set of rules represents the categorization of each number. Finally, rigorous experiments are carried out on 226 random Arabic numerals selected from 40 images of Iraqi car plate numbers. The proposed method attained an accuracy of 97%.

show abstract

Improved method of handwritten digit recognition tested on MNIST database

Cited by 177 publications

References 8 publications

Appropriateness of Dropout Layers and Allocation of Their 0.5 Rates across Convolutional Neural Networks for CIFAR-10, EEACL26, and NORB Datasets

Appropriateness of Dropout Layers and Allocation of Their 0.5 Rates across Convolutional Neural Networks for CIFAR-10, EEACL26, and NORB Datasets

Recognition of Hindi (Arabic) Handwritten Numerals

Off-Line Arabic (Indian) Numbers Recognition Using Expert System

Contact Info

Product

Resources

About