Module 4
Learn how agents learn through interaction, rewards, and experience using reinforcement learning techniques.
Discover how autonomous agents make optimal sequences of decisions through rewards and penalties. Learn Bellman equations, Q-tables, and Deep Q-Networks.
Understand the science of learning to act: how agents navigate dynamic worlds to maximize long-term cumulative reward.
Examine the components of an intelligent agent: the policy, the value function, and the internal environment model.
Understand the world the agent inhabits: deterministic vs stochastic worlds, full vs partial observability, and the Gymnasium API.
Understand the mathematical definition of State and the foundational Markov Property: the future depends only on the present.
Explore the action space: discrete choices (buttons, moves) versus continuous actuator controls (steering angles, voltages).
Master the engine of reinforcement: sparse vs dense rewards, reward shaping, and avoiding perverse alignment incentives.
Deep dive into the agent’s decision function: deterministic argmax policies versus stochastic probability distributions.
Master the fundamental trade-off: exploiting known high-value actions versus exploring uncertain actions to discover even greater rewards.
Learn how agents discover optimal strategies through Q-learning, a fundamental reinforcement learning algorithm.
Understand the Q-Table lookup structure: mapping discrete states (rows) and actions (columns) to estimated cumulative rewards.
Master the mathematical engine of Q-learning: TD error, learning rate alpha, discount factor gamma, and the Bellman optimality update.
Synthesize Deep Learning and Reinforcement Learning: Deep Q-Networks (DQN), Experience Replay, and Target Networks.