AI class overview: Difference between revisions
Jump to navigation
Jump to search
Line 50: | Line 50: | ||
* '''Benign''' vs '''Adversarial''' | * '''Benign''' vs '''Adversarial''' | ||
** Benign environments: not really out there to get you | ** Benign environments: not really out there to get you | ||
** Adversarial environments: harder to find good actions when your opponent is | ** Adversarial environments: harder to find good actions when your opponent is counteracting what you're trying to achieve |
Revision as of 20:26, 15 October 2011
This is an overview in point form of the content in the AI class.
Welcome to AI
Introduction
- Lots of work to do
- Will be ranked
Course Overview
Purpose of the class:
- To teach the basics of AI artificial intelligence
- To excite you
Structure:
- Videos -> Quizzes -> Answer Videos
- Assignments (like quizzes but without the answers) and exams
Intelligent Agents
- Agent has sensors to determine its environment
- Agent has actuators to affect its environment
- Agent interacts in a Perception-Action cycle
- Agent's Control Policy maps sensors to actuators
Applications of AI
- Finance
- Robotics
- Games
- Medicine
- Web
- Other fields
Terminology
- Fully Observable vs Partially Observable
- Fully Observable environment: if what your agent can sense at any point in time is completely sufficient to make the optimal decision, i.e. if the sensors can see the entire state
- Partially Observable environment: need memory in the agent to make the best possible decision, because the agents' sensors can't see the entire state
- When we look at Hidden Markov Models we will learn more about structuring memory when state is partially observable
- Deterministic vs Stochastic
- Deterministic environment: agent's actions uniquely determine the outcome, there is no randomness
- Stochastic environment: outcome of actions involve a certain level of randomness
- Discrete vs Continuous
- Discrete environment: finitely many states and actions
- Continuous environment: infinitely many states and actions
- Benign vs Adversarial
- Benign environments: not really out there to get you
- Adversarial environments: harder to find good actions when your opponent is counteracting what you're trying to achieve