Simple graphical model for graph-based semi-supervised learning

This project derives a simple graphical model for the objective function of the mincut approach.

For more detail your may refer to chapter 3 of my master thesis.

Note: Inference algorithms are mostly implemented as they are.

How to reproduce

• Install requirements.txt
• All test cases are implemented in source.py as individual functions.

Main modules

• data/... defines data processing scripts
• graph_construction constructs graph from input data, supports some simple graph construction methods
• ?_inference defines graph-based model inference.