EANN_TensorFlow

Produced by Jiaxin Zhang

This is the TensorFlow implementation for the paper <<EANN: Event Adversarial Neural Networks for Multi-Modal Fake News Detection>>, including industrial deployment.

Requirements

• python 3
• tensorflow == 1.14 (prefer to 1.14, >1.12 also satisfied)
• If encounter mismatch GPU Driver problem, please install tensorflow_gpu by conda install tensorflow-gpu=1.14

Run Example

# train
python train.py -m train
# package model for inference
python train.py -m package
# predict
python predict.py

Description

• data_new

The train, evaluate, test datas are saved here.

• process_data

Make data, no need to modify if only run the example.

• EANN.py

The main EANN model.

• config.py

Revise the configuration in this file.

• data_utils.py

The function provides data input for the model.

• model_helper.py

Supplenmentary functions for the main model.

• train.py

Function for training and packaging the model.

• predict.py

Function for prediction the model.

Reference

• reversal_gradient.py is from https://github.com/KnightZhang625/tf-dann/blob/master/flip_gradient.py
• textCNN revised from https://github.com/dennybritz/cnn-text-classification-tf/blob/master/text_cnn.py