A Novel Transfer Learning Approach upon Hindi, Arabic, and Bangla Numerals using Convolutional Neural Networks

@article{Tushar2017ANT,
  title={A Novel Transfer Learning Approach upon Hindi, Arabic, and Bangla Numerals using Convolutional Neural Networks},
  author={Abdul Kawsar Tushar and Akm Ashiquzzaman and Afia Afrin and Md. Rashedul Islam},
  journal={ArXiv},
  year={2017},
  volume={abs/1707.08385}
}
Increased accuracy in predictive models for handwritten character recognition will open up new frontiers for optical character recognition. [...] Key Method The model utilizes convolutional neural networks with backpropagation for error reduction and dropout for data overfitting. The output performance of the proposed neural network is shown to have closely matched other state-of-the-art methods using only a fraction of time used by the state-of-the-arts.Expand
Applying Data Augmentation to Handwritten Arabic Numeral Recognition Using Deep Learning Neural Networks
TLDR
A convolutional neural network model for recognizing handwritten numerals in Arabic language is proposed in this paper, where the dataset is subject to various augmentation in order to add robustness needed for deep learning approach. Expand
A Transfer Learning Approach for Handwritten Numeral Digit Recognition
TLDR
Using the transfer learning method for handwritten numeral digit recognition can significantly reduce the training time of the deep learning models, and slightly reduces the recognition accuracy. Expand
Handwritten Arabic numerals recognition using convolutional neural network
TLDR
A modification of previously proposed CNN architecture has given an accuracy of 98.91% and the proposed architecture has produced 99.76%, which is comparable to state-of-the-art results found in the domain of handwritten Arabic numeral recognition. Expand
An Efficient Recognition Method for Handwritten Arabic Numerals Using CNN with Data Augmentation and Dropout
TLDR
A convolutional neural network model for recognizing handwritten numerals in Arabic language is proposed in this paper, where the dataset is subject to various augmentations in order to add robustness needed for deep learning approach. Expand
An efficient method for improving classification accuracy of handwritten Bangla compound characters using DCNN with dropout and ELU
TLDR
This work proposes a novel method of recognition of compound characters in Bangla language using deep convolutional neural networks (DCNN) and efficient greedy layer-wise training approach and introduction of dropout technology mitigates data overfitting and Exponential Linear Unit (ELU) is introduced to tackle the vanishing gradient problem during training. Expand
Basic to Compound: A Novel Transfer Learning Approach for Bengali Handwritten Character Recognition
TLDR
This study shows that CNN trained on basic characters is well capable of recognizing compound characters with minimal retraining and performs better and also trains much faster than CNN fully trained on compound characters. Expand
Deep Learning Network Architecture based Kannada Handwritten Character Recognition
TLDR
A novel model for recognition of handwritten Kannada characters using transfer learning from Devanagari handwritten recognition system is presented, which has recorded an accuracy of 73.51% after evaluation in 10 epochs with VGG19 NET. Expand
OCR-Nets: Variants of Pre-trained CNN for Urdu Handwritten Character Recognition via Transfer Learning
TLDR
OCR-Nets, variants of (AlexNet & GoogleNet) for recognition of handwritten Urdu characters through transfer learning, and the experimental result shows that OCR-AlexNet and O CR-GoogleNet produce significant performance gains. Expand
A Novel Deep Convolutional Neural Network Architecture Based on Transfer Learning for Handwritten Urdu Character Recognition
TLDR
A novel deep convolutional neural network is proposed for handwritten Urdu character recognition by transfer learning three pre-trained CNN models which outperforms the individual CNNs and numerous conventional classification methods. Expand
Ink Analysis Using CNN-Based Transfer Learning to Detect Alteration in Handwritten Words
TLDR
An ink color based analysis using Convolutional Neural Network (CNN) automation method has been introduced for alteration detection using pre-trained AlexNet and VGG-16 architectures to study the effect of transfer learning on the problem at hand. Expand
...
1
2
...

References

SHOWING 1-10 OF 22 REFERENCES
Handwritten Arabic numeral recognition using deep learning neural networks
  • Akm Ashiquzzaman, A. Tushar
  • Computer Science
  • 2017 IEEE International Conference on Imaging, Vision & Pattern Recognition (icIVPR)
  • 2017
TLDR
A novel algorithm based on deep learning neural networks using appropriate activation function and regularization layer is proposed, which shows significantly improved accuracy compared to the existing Arabic numeral recognition methods. Expand
Transfer learning for Latin and Chinese characters with Deep Neural Networks
TLDR
This work analyzes transfer learning with Deep Neural Networks on various character recognition tasks and finds that pretrained nets consistently outperform randomly initialized nets on new tasks with few labeled data. Expand
CNN based common approach to handwritten character recognition of multiple scripts
TLDR
A convolutional neural network trained for a larger class recognition problem towards feature extraction of samples of several smaller class recognition problems of English, Devanagari, Bangla, Telugu and Oriya each of which is an official Indian script. Expand
Backpropagation Applied to Handwritten Zip Code Recognition
TLDR
This paper demonstrates how constraints from the task domain can be integrated into a backpropagation network through the architecture of the network, successfully applied to the recognition of handwritten zip code digits provided by the U.S. Postal Service. Expand
Dropout: a simple way to prevent neural networks from overfitting
TLDR
It is shown that dropout improves the performance of neural networks on supervised learning tasks in vision, speech recognition, document classification and computational biology, obtaining state-of-the-art results on many benchmark data sets. Expand
Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)
TLDR
The "exponential linear unit" (ELU) which speeds up learning in deep neural networks and leads to higher classification accuracies and significantly better generalization performance than ReLUs and LReLUs on networks with more than 5 layers. Expand
A New Text Categorization Technique Using Distributional Clustering and Learning Logic
TLDR
A new text categorization method is presented that combines the distributional clustering of words and a learning logic technique, called Lsquare, for constructing text classifiers, that achieves higher or comparable classification accuracy and F1 results compared with SVM on exact experimental settings with a small number of training documents. Expand
Self-taught learning: transfer learning from unlabeled data
TLDR
An approach to self-taught learning that uses sparse coding to construct higher-level features using the unlabeled data to form a succinct input representation and significantly improve classification performance. Expand
Heterogeneous Transfer Learning for Image Clustering via the SocialWeb
TLDR
Experiments show that the approach in heterogeneous transfer learning based on the auxiliary data is indeed effective and promising, and PLSA is extended to help transfer the knowledge from social Web data, which have mixed feature representations. Expand
Can chinese web pages be classified with english data source?
TLDR
This paper proposes an information bottleneck based approach to address the cross-language classification problem of Chinese and English Web pages, and significantly improves several existing supervised and semi-supervised classifiers. Expand
...
1
2
3
...