Configurations

• Change HOME_PATH in CONFIG.py as the current path

Data Prepare

CENSINCOME

• Download data
• Put census-income.data and census-income.test in Data/Downloads/census-income
• Run DataProcess/CENSINCOME/process.py

COVTYPE

• Download data
• Put covtype.data in Data/Downloads/COVTYPE/
• Run DataProcess/COVTYPE/process.py

IJCAI18X

• Download data
• Generate train.csv and test.csv with xgboost1.py
• Put train.csv and test.cvs in Data/RawData/IJCAI18X
• Run DataProcess/IJCAI18X/process.py

KDDCUP19

• Download data
• Put train_queries.csv, train_plans.csv, train_clicks.csv, profiles.csv in Data/Downloads/KDDCUP19P1
• Run DataProcess/KDC/process.py

Training

Train and save the model

• Use the scripts in Train/{DATASET}/
• Will train the model and save the model in Data/Saved/{DATASET}/[VOTERS or DNN]/model

Explanation

Generate intermediate results of voting analysis

• Run SaveInter/save.py

Visualizations for explanation

• Run E1 to E6 in Explain/analysis, the visualizations will be saved in Explain/out

Demo system for local/global decision path visualization

• Open the path of the django project in Site/
• Start the server: python manage.py runserver 0.0.0.0:8000
• Use the system in browser with URL: 127.0.0.1:8000/{global/local}/{covtyp/kdc/ijcai/cens}

Citation

If you use this code, please cite our paper "Discerning Decision-Making Process of Deep Neural Networks with Hierarchical Voting Transformation" published in NeurIPS 2021.