Linear Bandit Implementation

Python3 based on https://github.com/eraser347/SqueezeAll.git

Directory tree

├────── results
├── data.py
├── evaluation.py
├── exp.py
├── regrets.py
└── train.py

• data.py contains the codes for generating contexts.
• evaluation.py evaluates the four algorithms in the paper.
• exp.py to generates the results and save it to ./results.
• regrets.py plots the cumulative regrets with the data saved in ./results
• To obtain the results, enter python3 exp.py and then run python3 regrets.py
• exp.ipynb for simulation in notebook.

Requirements

• python 3
• numpy
• scipy
• matplotlib
• tqdm