SPN-Z-SL

Structure Learning for Continuous Variables in SPN-Z

Used for all experiments in Online Structure Learning for Sum-Product Networks Created by Wilson Hsu and Agastya Kalra.

Dependencies:

• Scipy

• Numpy

• Python 3.3+

Usage:

To run any of the experiments in the dataset first create a folder in the library directory called SPN-Z-SL/data/dataset_name/.

Then create 10 files of the dataset as follows:

dataset_name.1.data ... dataset_name.10.data.

then add ~/path/to/SPN-Z-SL/ to your PYTHONPATH.

finally in experiments/read_data.py, add the line to run an experiment. The file explains the parameters.

then in the libraries main directory type:

python3 experiment/real_data.py

REAL_NVP Experiments: https://github.com/KalraA/real-nvp/