Implementation of FLIPHAT algorithm for Jointly Differentially Private Sparse Linear Contextual Bandits - see paper (Sunrit Chakraborty, Saptarshi Roy, Debabrota Basu)

The main codes are in SparseBandit (see example.ipynb for a brief demo). Check plots folder to see the figures generated from simulation studies (for more details, see paper).

Packages used (Python 3.9.18):

1. jax (0.4.23)
2. numpy (1.26.3)
3. sklearn (1.3.0), scipy (1.11.4) - for sparsity agnostic bandit
4. matplotlib (3.8.4), tqdm (4.66.2) - for graphics