samsonq / DeepMAB

MAB and linear/non-linear Contextual MAB algorithms

Deep Contextual MAB

MAB and linear/non-linear Contextual MAB algorithms.

Algorithms

Multi-Arm Bandits

• Epsilon Greedy
• UCB
• Thompson Sampling

Contextual Multi-Arm Bandits

• Neural Net Epsilon Greedy
• LinUCB
• Neural Net UCB

Usage Instructions

• This project is published on PyPI. To install package, run:

  pip install deep-mab
  

• To run the algorithms, import the package and call the respective functions. For example, to run the LinUCB algorithm, run:

  from deep_mab.cmab import LinUCB
  model = LinUCB(n_arms=10, alpha=1, fit_intercept=True)
  model.fit(X_train, y_train)
  model.predict(X_test)
  

• For more details, refer to the documentation.

• To run the examples, clone the repository and run the following commands:

  cd deep-mab
  pip install -r requirements.txt
  python examples/linucb_example.py
  

• To run the tests, run the following commands:

  cd deep-mab
  pip install -r requirements.txt
  pytest