Untitled Flashcards Set

Reinforcement Learning (RL)

Teaching a machine to learn by trial and error, using rewards (good) and penalties (bad) to improve decisions.

Q-Learning

An RL method where an agent learns a Q-table to decide the best actions in each situation.

Doing RL by Hand

Manually updating Q-values by picking actions (random/best), observing rewards, and repeating to improve decisions.

Bandits

A scenario where the agent chooses between options (like slot machines) to find the best reward.

Multi-Armed Bandits

Choosing between multiple options (e.g., slot machines), balancing exploration and exploitation.

Contextual Bandits

Bandits with context where the choice depends on the situation.

Epsilon-Greedy Strategy

A method to balance exploration and exploitation by exploring randomly with chance ε and exploiting the best-known action otherwise.

Upper Confidence Bound (UCB)

A strategy for balancing good rewards and uncertain actions.

Thompson Sampling

Uses probabilities to pick actions with the highest potential reward, balancing exploration and exploitation naturally.