It is a start up tutorial for tensorflow.
Tips: this tutorial can be successfully run in windows 10 system.
Environment required: windows 10 system (maybe ubuntu 16.04) anaconda 3 (easy to install python3.6) pycharm (convenient for student to activate) library function -> tensorflow, numpy
Thanks to MOVAN python tutorial's video's support!
[test5-BasicNN.py]: It is a basic 2 layers neural network to learn a regression function for f(x) = square(x) + 0.5
[test8-MNIST1layer.py]: It is a 1 layer neural network to learn the MNIST handwritten numeral dataset.
[test8.1-MNIST2layers.py]: It is a 2 layers neural network to learn the MNIST handwritten numeral dataset.
[test9-MNISTCnn.py]: It is a basic convolutional neural network to learn the MNIST handwritten numeral dataset.
[test10-SavePara.py]: It is a demo of saving parameter of network.
[test10.1-RestorePara.py]: It is a demo of restoring parameter of network.(we can only save the parameters but cannot save save the construction)
[test11-RNN.py]: It is a demo of MNIST recognition by LSTM model.
[test14-CLSTM.py]: It is a demo of MNIST recognition by CLSTM(CNN+LSTM) model. PS: Some annotations are in Chinese.