/spn-opt-discrim

Adaptation of SPN parameter learning by Zhao et al. to support classification under missing data setting.

Primary LanguageC++OtherNOASSERTION

Adaptation of "Parameter Learning of Sum-Product Networks" for learning models for missing data classification

This contains the original code implementing the methods described in:

  • A Unified Approach for Learning the Parameters of Sum-Product Networks by H. Zhao, P. Poupart and G. Gordon, NIPS 2016.
  • Collapsed Variational Inference for Sum-Product Networks by H. Zhao, T. Adel, G. Gordon and B. Amos, ICML 2016.

Along with some code to enable classification with missing data (see here).

The original README can be found here

Follow original instructions to build the source files.