Description: 基于神经网络的手写数字识别的源代码，绝对能够正常编译并运行！-based on neural network handwritten numeral recognition of the source code is absolutely normal to compile and run! Platform: |
Size: 207872 |
Author:田巾 |
Hits:
Description: 闲时无聊，搭了一个基于深度神经网络的手写数字识别系统。该系统在手写数字数据库mnist测试达到了99.22 的准确率。整个系统基于C++开发，可以很方便的移植到其他平台。
其中手写数字数据库mnist（http://yann.lecun.com/exdb/mnist/），有60000个训练样本数据集和10000个测试用例。它是由Google实验室的Corinna Cortes和纽约大学柯朗研究所的Yann LeCun建立的一个手写数字数据库。同时它是nist数据库的一个子集。
我们这里给出了训练和测试的tools以及一个demo。打开解压包，有两个文件夹，一个是“训练和测试的tools”，一个是“demo”。
详细的使用方法参看我的博客：http://blog.sciencenet.cn/home.php?mod=space&uid=907554&do=blog&id=772908-Leisure bored, take a handwriting recognition system based on neural network depth. The system is tested in handwritten digital database mnist reached 99.22 accuracy rate. The whole system is based on C++ development, can be easily ported to other platforms.
Where handwritten digital database mnist (http://yann.lecun.com/exdb/mnist/), there are 60,000 training sample data sets and 10,000 test cases. It is composed of Yann Google Labs Corinna Cortes and New York University s Courant Institute of LeCun establish a handwritten digital database. Also it is a subset of the database nist.
Here are the tools of our training and testing as well as a demo. Open unzip the package, there are two folders, one is "training and testing tools", one is "demo".
Detailed usage see my blog: http://blog.sciencenet.cn/home.php?mod=space&uid=907554&do=blog&id=772908 Platform: |
Size: 28878848 |
Author:Shi |
Hits:
Search - digital recognition neural network demo