Bandits

The bandit.py module includes several simple multi-arm bandit environments.

The policies.py module implements a number of standard multi-arm bandit policies.

1. Bandits

   • MAB: Bernoulli, Multinomial, and Gaussian payout distributions
   • Contextual MAB: Linear contextual bandits

2. Policies

   • Epsilon-greedy
   • UCB1 (Auer, Cesa-Bianchi, & Fisher, 2002)
   • Conjugate Thompson sampler for Bernoulli bandits (Thompson, 1933; Chapelle & Li, 2010)
   • LinUCB (Li, Chu, Langford, & Schapire, 2010)

Plots

<p align="center"> </p>