Handwritten Character Recognition Based on BP Neural Network
-
摘要: 文章对比了使用TensorFlow搭建的两种使用MNIST数据集训练的神经网络模型,并对神经元个数、激活函数、学 习率、优化器等参数不断优化。结果表明,在足够的训练数据集下,简单的神经网络模型对手写字符的识别已经能达到较高 水平,额外的技巧则大大增强了模型的泛化能力与鲁棒性。
-
关键词:
- 手写字符 /
- 识别 /
- 神经网络 /
- MNIST /
- TensorFlow
Abstract: The article compares twoneural network models trained using TensorFlow on MNIST datasets, and optimizes parameters such as the number of neurons, activation function, learning rate, and optimizer. The results show that the simple neural network model can achieve high performance on handwriting recognition under sufficient training data sets, and the additional technical optimization greatly enhances the generalization ability and robustness of the model.-
Key words:
- Handwritten characters /
- Recognition /
- Neural network /
- MNIST /
- TensorFlow
-
[1] 李丹.基于 BP 神经网络的多样本手写体字符识别[J].软件,2016,37(7):103-108. [2] 许宜申,顾济华,陶智,等.基于改进BP神经网络的手写字符识别[J].通信技术,2011,44(5):106-109+118. [3] 丁蒙,戴曙光,于恒.卷积神经网络在手写字符识别中的应用[J].软件导刊,2020,19(1):275-279. [4] 黎育权.基于集成学习改进的卷积神经网络的手写字符识别[J].电子技术与软件工程,2018(9):167. -

计量
- 文章访问数: 426
- HTML全文浏览量: 64
- PDF下载量: 45
- 被引次数: 0