Bandit Learning Research

Multi-armed bandit algorithms; Research done at Dr Ji's SNAIL lab @ Virginia Tech

Implements different bandit algorithms that solve the exploration vs exploitation problem.

Algorithms

• Epsilon-greedy: Random exploration with probability ε
• UCB: Upper confidence bound approach
• Thompson Sampling: Bayesian method
• Contextual bandits: Uses additional info to make decisions

Requirements

numpy, matplotlib, scipy