- 期刊
A Design and Implementation of Handwritten Digit Recognition Based on DBN
摘要
The principle of deep belief network is to combine the low-level feature combination with the higher-level feature combination, and the unsupervised learning method reduces the amount of personnel labor. The main method of using DBN for handwritten digit recognition is studied. Using the pictures in the data for training, the accuracy is as high as 93.42%. Higher than the SVM under the same conditions. In addition, the introduction of the Dropout parameter in the deep learning network can achieve higher recognition accuracy with a small number of samples.
延伸閱讀
- 程宏琪（2010）。Implementation of Hand-Shape Recognition〔碩士論文，國立暨南國際大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0020-2208201021024200
- Cho, H. F. (2018). On-Pen Handwritten Word Recognition Using Long Short-Term Memory Model [master's thesis, National Tsing Hua University]. Airiti Library. https://www.airitilibrary.com/Article/Detail?DocID=U0016-1803201914433641
- 熊星皓（2015）。Handwriting beautification based on analysis of smartphone sensor data〔碩士論文，國立交通大學〕。華藝線上圖書館。https://doi.org/10.6842/NCTU.2015.00102
- Hoi, C. C. (2010). A Study on Handwritten Character Recognition [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU.2010.10634
- Pang, S., & Yang, X. (2016). Deep Convolutional Extreme Learning Machine and Its Application in Handwritten Digit Classification. Computational Intelligence and Neuroscience, 2016(), 549-558-044. https://doi.org/10.1155/2016/3049632