PrivPfC: differentially private data publication for classification

PrivPfC is a differentially private algorithm for releasing data for classification analysis. The key idea underlying PrivPfC is to privately select, in a single step, a grid, which partitions the data domain into a number of cells. This selection is done by using the exponential mechanism with a novel quality function, which maximizes the expected number of correctly classified records by a histogram classifier. PrivPfC supports both the binary classification and the multiclass classification.

For more details, please see our paper: Dong Su, Jianneng Cao, Ninghui Li, Min Lyu: PrivPfC: differentially private data publication for classification. VLDB J. 27(2): 201-223 (2018).

How to run

• cd ./src/single_grid
• To publish the adult data for binary classification on the income attribute,
  • cd ./src/experiment_basic.py
  • python experiments_basic.py --dataset_name=adult --epsilon=1.0 --fold_num=0 --pool_size_threshold=10000
• To publish the adult data for 3-class classification on the marital-status attribute,
  • cd ./src/experiment_basic.py
  • python experiments_basic.py --dataset_name=adult_marital-status --epsilon=1.0 --fold_num=0 --pool_size_threshold=10000

Environment requirements

• Python 2.7.12
• Numpy 1.12.1
• Pandas 0.19.1

Maintainer: Dong Su, [email protected]