An implementation of the Maximum Entropy Generative model based as discussed in the paper A high-bias, low-variance introduction to Machine Learning for physicists (or here on arXiv) can be found in gempy.maximum_entropy_model.py
, slides are located in doc
.
In gempy.mnist.mnist_generator.py
we apply the Maximum Entropy Principle to generate (or fit) an Ising-based generative model on single numbers in the mnist-dataset. examples/mnist/mnist_generator.ipynb
provides a classifier cnn as well as MNISTGenerator
instances which can be trained to generate numbers from 0
to 9
using the Maximum Entropy Principle, which can then be evaluated with the classifier network. Here are some example figures from the jupyter-notebook: