RNN_for_Handwriting_Recognition-master
所属分类：人工智能/神经网络/深度学习
开发工具：Python
文件大小：12059KB
下载次数：4
上传日期：2019-06-16 17:10:10
上 传 者：
遗传算法
说明： RNN手写识别神经网络，采用MNIST数据集，效果较好
(RNN handwriting recognition neural network, using MNIST data set, has better effect.)
文件列表:
MNIST_data (0, 2019-01-11)
MNIST_data\t10k-images.idx3-ubyte (7840016, 2019-01-11)
MNIST_data\t10k-labels.idx1-ubyte (10008, 2019-01-11)
MNIST_data\train-images.idx3-ubyte (47040016, 2019-01-11)
MNIST_data\train-labels.idx1-ubyte (60008, 2019-01-11)
images (0, 2019-01-11)
images\BackPropagations.png (691892, 2019-01-11)
images\BatchError.png (87697, 2019-01-11)
images\Results.png (69276, 2019-01-11)
images\TestingResults.png (112398, 2019-01-11)
nn_for_MNIST.py (8392, 2019-01-11)
## NN for handwriting recognition from pure numpy
## Steps
Step1. reading the images and the labels
Step2. Setting learning rate, dropout fraction, epoches and batchs
Step3. Setting activation function (Sigmoid)
Step4. Calculating the nn fowardings
Step5. Calculating the nn back propogations
Step5, Refresh nn and the gradients
Step6. Stastistics of accuracy and batch regressions
## Test results
近期下载者：
相关文件：
收藏者：