Handwritten digits classification from MNIST with TensorFlow
本例使用python3.6开发
本例依赖 tensorflow 1.1.0 | pillow 4.1.1
如果本例对您有帮助,请不要吝惜您的star!
本例训练了4种分类器:
-
无隐藏层,激活函数为softmax
-
5个隐藏层,激活函数为sigmoid
-
5个隐藏层,激活函数为relu
-
2个 卷基层+池化层,1个全连接层
pip install -r requirements.txt
python app.py
python train.py
按照如下选择需要训练的分类器
Please Input the trainer:
-------------------------
a: softmax
b: sigmoid_5_layers
c: relu_5_layers
d: conv2d
-------------------------
trainer: (a)