mnist_training_examples

MNIST Dataset Training Examples

1. Simple MNIST classification with logistic regression. It's a practice to learn the MNIST dataset and use a simple regressor to classify the dataset.
2. MNIST Training, saving and loading model. It's a practice to learn how to save and load models trained by tensorflow. Reference: 10_save_restore_net.py.
3. Visualize training process. A practice to learn how to visualize the training process with tensorboard. Reference: mnist_with_summaries.py.
4. Train MNIST dataset via CNN. A simple convolutional neural networks example to deal with a classification tasks. Reference: 05_convolutional_net.py.
5. Train MNIST dataset via RNN. A simple recurrent neural networks example to train MNIST dataset. Reference: recurrent_network.py, Supervised Sequence Labelling with Recurrent Neural Networks.
6. Train MNIST with autoencoder. An example of unsupervised learning. Reference: Autoencoders and Sparsity.
7. Train MNIST to explore the effect of training epochs for losses in training dataset and in validation dataset (Maxout Net). (Deep Learning Book P152 Figure)
8. Example to convert MNIST dataset to TFRecords. link
9. Maxout Networks in Tensorflow to train MNIST dataset. Reference: philipperemy/tensorflow-maxout, Maxout Networks.

Reference book: Tensorflow 技术解析与实战, 深度学习 (Deep Learning).