Expectation-Maximization (EM) for mixture of Multinomials

This Python code implements the EM algorithm for the mixture of Multinomials. The generative process that explains the mixture of Multinomials: For each document, we sample a topic from a document-specific multinomial. Given this topic, we sample a word from the Multinomial distribution that represents this topic.

Running the code

To run the code with a synthetic dataset

python -i mix_multi.py