TensorFlow Tutorial - Ensemble Learning

Notes by Magnus Erik Hvass Pedersen: https://github.com/Hvass-Labs/TensorFlow-Tutorials/blob/master/05_Ensemb...

This tutorial shows how to use a so-called ensemble of convolutional neural networks. Instead of using a single neural network, we use several neural networks and average their outputs.

This is used on the MNIST data-set for recognizing hand-written digits. The ensemble improves the classification accuracy slightly on the test-set, but the difference is so small that it is possibly random. Furthermore, the ensemble mis-classifies some images that are correctly classified by some of the individual networks.

This tutorial builds on the previous tutorials, so you should have a basic understanding of TensorFlow and the add-on package Pretty Tensor. A lot of the source-code and text here is similar to the previous tutorials and may be read quickly if you have recently read the previous tutorials.