This repo is an exploratory codebase for learning and interacting with reinforcement learning concepts, libraries, and notebooks.
The goal of this project is to learn through practice by implementing a variety of reinforcement learning algorithms. The project can be broken up into two sections: algorithms, and integrations. Where integrations will be interfaces with other packages that provide test environments for RL.
- Action-Value
- Stationary k-armed Bandit Problems
- Non-stationary k-armed Bandit Problems
- greedy
- ε-greedy
- Optimistic Initial Values
- Upper-Confidence-Bound (UCB)
- Bandit Gradient Algorithm
- Associative Search (Contextual Bandits)
- argmax
- k-armed Bandit