By: Archana Vyas, Vedanshi Pandya, and Yashvi Desai
Handwritten digit recognition is a fundamental task in the field of pattern recognition and computer vision with numerous applications. The development of deep learning algorithms has greatly increased the efficiency and accuracy of recognition of handwritten digits systems. However, the choice of optimizer plays a crucial role in determining the convergence rate and final performance of deep learning models. In this study, we provide a thorough analysis of many deep learning optimizers for the task of handwriting digit recognition in Gujarati. The study focuses on optimizing the recognition accuracy and convergence speed of convolutional neural network (CNN) models trained on a dataset comprising handwritten digits in the Gujarati script. The performance of popular optimizers such as Stochastic Gradient Descent (SGD), Adam, RMSprop, Adagrad and Adadelta are compared and analyzed. The evaluation metrics include classification accuracy, training time, and convergence behavior. The insights gained from this study can guide the selection of optimizers for similar tasks in other languages and domains, facilitating the development of efficient and accurate deep learning models.
Keywords: Deep Learning, Gujarati Numerals, Handwritten script recognition, CNN, Gradient descent, RMSprop, Stochastic Gradient Descent (SGD)
Citation:
Refrences:
- Ghosh MM, Maghari AY. A comparative study on handwriting digit recognition using neural networks. In2017 international conference on promising electronic technologies (ICPET) 2017 Oct 16 (pp. 77-81). IEEE.
- Verma P, Tripathi V, Pant B. Comparison of different optimizers implemented on the deep learning architectures for COVID-19 classification. Materials Today: Proceedings. 2021 Jan 1;46:11098-102.
- Desai AA. Gujarati handwritten numeral optical character reorganization through neural network. Pattern recognition. 2010 Jul 1;43(7):2582-9.
- Postalcıoğlu S. Performance analysis of different optimizers for deep learning-based image recognition. International Journal of Pattern Recognition and Artificial Intelligence. 2020 Feb 14;34(02):2051003.
- Patel C, Desai A. Extraction of characters and modifiers from handwritten Gujarati words. International Journal of Computer Applications. 2013 Jan 1;73(3).
- MJ B, Kale KV. Gujarati Numeral Recognition: Affine Invariant Moments Approach. 1st International Conference on Recent Trends in Engineering & Technology. 2012. Special Issue of International Journal of electronics, Communication & Soft Computing Science & Engineering. 140-146p.
- Ahlawat S, Choudhary A, Nayyar A, Singh S, Yoon B. Improved handwritten digit recognition using convolutional neural networks (CNN). Sensors. 2020 Jun 12;20(12):3344.
- Acharya S, Pant AK, Gyawali PK. Deep learning based large scale handwritten Devanagari character recognition. In2015 9th International conference on software, knowledge, information management and applications (SKIMA) 2015 Dec 15 (pp. 1-6). IEEE.
- Vaidya R, Trivedi D, Satra S, Pimpale M. Handwritten character recognition using deep-learning. In2018 second international conference on inventive communication and computational technologies (ICICCT) 2018 Apr 20 (pp. 772-775). IEEE.
- Setiowati S, Franita EL, Ardiyanto I. A review of optimization method in face recognition: Comparison deep learning and non-deep learning methods. In2017 9th International Conference on Information Technology and Electrical Engineering (ICITEE) 2017 Oct 12 (pp. 1-6). IEEE.
- Hallale SB, Salunke GD. Offline handwritten digit recognition using neural network. International Journal of Advanced Research in Electrical, Electronics and Instrumentation Engineering. 2013 Sep;2(9):4373-7.
- Maloo M, Kale KV. Support vector machine based Gujarati numeral recognition. International Journal on Computer Science and Engineering. 2011 Jul;3(7):2595-600.
