/pmln

Pairwise Multi-Layer Nets for Learning Distributed Representation of Multi-field Categorical Data

Primary LanguagePythonMIT LicenseMIT

Pairwise Multi-Layer Nets for Learning Distributed Representation of Multi-field Categorical Data

This repo aims to provide an algorithm implementation for Pairwise Multi-Layer Nets (PMLN) for Learning Distributed Representation of Multi-field Categorical Data and its baselines.

Installation

  1. Intsall the dependencies.
sudo pip install keras, tensorflow

Note: The code was wrote in 2016, which means it requires (very old version) TensorFlow==0.10.0 and Keras==1.0.0.

  1. Preprae data according to the instructions in data folder.

Runing Experiments

python ctr_distcrimination.py % for FMLN-1
python ctr_distcrimination_concat.py % for FMLN-2