Harflerden Oluşan Genişletilmiş MNİST Veri Kümesinin Derin Öğrenme Tabanlı Tasarlanmış Sinir Ağı Modeli ile Sınıflandırılması

Year 2021, , 681 - 690, 30.09.2021

Mesut Toğaçar

https://doi.org/10.21605/cukurovaumfd.1005374

Abstract

MNİST veri kümesi, model öğrenmesi, görüntü işleme, sınıflandırma süreçlerinin gerçekleştirilmesinde standart bir ölçüt olarak kullanılmaktadır. MNİST veri kümesi içerisinde; el yazısı formatında hazırlanmış büyük harf, küçük harf ve rakam görüntülerinden oluşmaktadır. Genişletilmiş MNİST veri kümesi, MNİST veri kümesi ile aynı değerler dizisi formatında hazırlanmış daha kapsamlı ve sınıflandırma süreci daha zor bir veri seti türüdür. Günümüzde birçok alanda yapay zekâ tabanlı çalışmalar ilgi görmeye başlamıştır. Bu çalışmada, Genişletilmiş MNİST veri kümesinin eğitilmesi ve sınıflandırması amacıyla Python dilinde tasarlanmış yeni bir sinir ağı modeli önerilmektedir. Önerilen modelde, ön işlem adımı olarak veri büyütme
yöntemi eğitim verileri için uygulanmıştır ve 26 harf kategorik olarak sınıflandırıldı. Sınıflandırma sürecinde genel doğruluk başarısı %94,73 olarak elde edildi. Önerdiğimiz model, el yazısı görüntülerinin sınıflandırılmasında başarılı bir analiz gerçekleştirdiği gözlemlendi

Keywords

Derin öğrenme, Genişletilmiş MNIST veri kümesi, Harf sınıflandırması, Sinir ağları

Classification of Extended MNIST Dataset Consisting of Letters with Deep Learning-Based Designed Neural Network Model

Abstract

MNİST dataset is used as a standard criterion for model learning, image processing and classification processes. In the MNIST dataset; it consists of uppercase, lowercase letters and numbers in handwriting format. The expanded MNIST data set is a more comprehensive type of data set prepared in the same paradigm format as the MNIST dataset, and the classification process is more difficult. Today, artificial intelligence-based studies in many areas have started to attract attention. In this study, a new neural network model designed in Python language is proposed in order to train and classify the extended MNIST dataset. In the proposed model, data enlargement method as a preprocess step was applied for training data and 26 letters were categorically classified. The overall accuracy success achieved in the classification process was %94.73. The proposed model we are observed to perform a successful analysis in classifying handwritten images.

Keywords

Deep learning, Extended MNIST dataset, Letters classification, Neural networks

References

• 1. Lundervold, A.S., Lundervold, A., 2019. An Overview of Deep Learning in Medical Imagingm Focusing on MRI. Z Med Phys. 29(2), 102–127. https://doi.org/10.1016/j.zemedi.2018.11.002.
• 2. AlQuraishi, M., 2019. ProteinNet: a Standardized Data Set for Machine Learning of Protein Structure. BMC Bioinformatics, 20,311. https://doi.org/10.1186/s12859-019-2932-0.
• 3. Zhang, J.M., Harman, M., Ma, L., Liu, Y., 2020. Machine Learning Testing: Survey, Landscapes and Horizons. IEEE Trans Softw Eng. 99, 1–1. https://doi.org/10.1109/tse.2019.2962027.
• 4. Studer, L., Alberti, M., Pondenkandath, V., Goktepe, P., Kolonko, T., Fischer, A., Liwicki, M., Ingold, R., 2019. A Comprehensive Study of Imagenet Pre-training for Historical Document Image Analysis. Proc Int Conf Doc Anal Recognition, ICDAR. 720–725. https://doi.org/10.1109/ICDAR.2019.00120.
• 5. Wei, Z., Wang, F., 2019. Adaptive Cascade Single-shot Detector on Wireless Sensor Networks. EURASIP J Wirel Commun Netw. 150. https://doi.org/10.1186/s13638-019-1440-2.
• 6. Lecun, Y., MNIST handwritten digit database, Corinna Cortes and Chris Burges. http://yann.lecun.com/exdb/mnist/. Erişim tarihi: 7 Haziran 2021.
• 7. Cohen, G., Afshar, S., Tapson, J., Van Schaik, A., 2017. EMNIST: Extending MNIST to Handwritten Letters. Proc Int Jt Conf Neural Networks 2017-May:2921–2926. https://doi.org/10.1109/IJCNN.2017.7966217
• 8. Jiang W., 2020. MNIST-MIX: A Multi-Language Handwritten Digit Recognition Dataset. IOP SciNotes 1:025002. https://doi.org/10.1088/2633-1357/abad0e
• 9. Amelia, A., 2018. Convolution Neural Network to Solve Letter Recognition Problem.
• 10.Grzelak, D., Podlaski, K., Wiatrowski, G., 2019. Analyze the Effectiveness of an Algorithm for Identifying Polish Characters in Handwriting Based on Neural Machine Learning Technologies. J King Saud Univ.- Comput Inf. Sci. https://doi.org/10.1016/j.jksuci.2019.08.001.
• 11.Jayasundara V., Jayasekara S., Jayasekara, H., Jayasekara, H., Rajasegaran, J., Seneviratne, S., Rodrigo, R., 2019. TextCaps: Handwritten Character Recognition with Very Small Datasets. In: 2019 IEEE Winter Conference on Applications of Computer Vision (WACV). 254–262.
• 12. Vaila, R., Chiasson, J., Saxena, V., 2020. A Deep Unsupervised Feature Learning Spiking Neural Network with Binarized Classification Layers for EMNIST Classification using Spyke Flow. 2002.11843.
• 13.Crawford C., 2017. Extended MNIST Letter Dataset. In: Kaggle. https://www.kaggle.com/crawford/emnist. Erişim tarihi: 7 Haziran 2021.
• 14. Shorten, C., Khoshgoftaar, T.M., 2019. A Survey on Image Data Augmentation for Deep Learning. J Big Data. 6, 60. https://doi.org/10.1186/s40537-019-0197-0.
• 15.Image Data Augmentation. In: Keras Blog. https://keras.io/api/preprocessing/image/. Erişim tarihi: 6 Haziran 2021
• 16. Dokuz, Y., Tufekci, Z., 2021. Mini-batch Sample Selection Strategies for Deep Learning Based Speech Recognition. Appl Acoust 171:107573. https://doi.org/10.1016/j.apacoust.2020.107573
• 17. Gadekallu, T.R., Khare, N., Bhattacharya, S., Singh, S., Maddikunta, P.K.R., Srivastava, G., 2020. Deep Neural Networks to Predict Diabetic Retinopathy. J Ambient Intell Humaniz Comput. https://doi.org/10.1007/s12652-020-01963-7.
• 18. Suárez-Paniagua, V., Segura-Bedmar, I., 2018. Evaluation of Pooling Operations in Convolutional Architectures for Drug-drug Interaction Extraction. BMC Bioinformatics 19(S8), 209. https://doi.org/10.1186/s12859-018-2195-1.
• 19. Lu, J., Ye, Y., Xu, X., Li, Q., 2019. Application Research of Convolution Neural Network in Image Classification of Icing Monitoring in Power Grid. EURASIP J Image Video Process. 49, 1-11. https://doi.org/10.1186/s13640-019-0439-2.
• 20. Luo, Y., Wong, Y., Kankanhalli, M., Zhao, Q., 2020. Softmax: Improving Intraclass Compactness and Interclass Separability of Features. IEEE Trans Neural Networks Learn Syst. 31, 685–699. https://doi.org/10.1109/tnnls.2019.2909737
• 21. Zhang, X., Zou, Y., Shi, W., 2017. Dilated Convolution Neural Network with LeakyReLU for Environmental Sound Classification. In: 2017 22nd International Conference on Digital Signal Processing (DSP). 1–5.
• 22. Srivastava, N., Hinton, G., Krizhevsky, A., Sutskever, I., Salakhutdinov, R., 2014. Dropout: A Simple Way to Prevent Neural Networks from Overfitting. J Mach Learn Res 15(1), 1929-1958.
• 23. Toğaçar, M., Ergen, B., Sertkaya, M.E., 2019. Zatürre Hastalığının Derin Öğrenme Modeli il Tespiti. Fırat Üniversitesi Mühendislik Bilim Dergisi, 31(1), 223–230.
• 24. Sertkaya, M.E., Ergen, B., Togacar, M., 2019. Diagnosis of Eye Retinal Diseases Based on Convolutional Neural Networks Using Optical Coherence Images. In: 2019 23rd International Conference Electronics. 1–5