/Handwritten_Digits_Recognition

🈚️TF/PT框架的手写数字识别,通过ndarray搭建神经网络中的权重矩阵,precision rate: 98%

Primary LanguagePython

Handwritten_Digits_Recognition

First, run python3 mnist_csv3.py in terminal.

Then you will see four .csv files which are used for training and testing.

Finally, run python3 NeuralNetwork.py train_image.csv train_label.csv test_image.csv

You will see a new file called test_predictions.csv.

The accuracy for training can be 98% and testing accuracy can be 93%-95%