摘要
基于谷歌第二代人工智能学习系统Tensor Flow,构建BP神经网络模型。将手写字符作为训练集输入神经网络,训练过程中不断调整权值和阈值,最终得到有较高识别精度的模型。体现了Tensor Flow在提高建模、编程、分析效率中的作用。通过此开发流程介绍,为进一步使用Tensor Flow构建复杂神经网络提供了参考。
Based on Google＇s second generation of artificial intelligence learning system--TensorFlow, build a BP neural net- work model. Use handwritten characters as training set of inputs of neural network, constantly revise weight value and threshold value in the process of training, and get a higher identification precision of the model. It Embodies the TensorFlow＇ s effect in improving the efficiency of modeling, programming and analysis. Through introducing the development process, It provides ref- erence which use TensorFlow building complex neural networks.
作者
张俊
李鑫
ZHANG Jun, LI Xin （College of Information Engineering, Taiyuan University of Technology, Taiyuan 030024, China）
出处
《电脑知识与技术》
2016年第6期199-201,共3页
Computer Knowledge and Technology
基金
基金项目：太原理工大学国家级大学生创新创业项目资助（编号201410112013）
山西省大学生创新创业项目资助（编号2014053）
