Ai 02 intelligent_agents(1)

1
Artificial Intelligence
Chapter 2: Intelligent Agents
Andreas Zell
After the Textbook: Artificial IntelligenceAfter the Textbook: Artificial Intelligence,
A Modern Approach
by Stuart Russel and Peter Norvig (3rd Edition)
• Rational agent: one that behaves as well as
possible
This behaviour depends on the environment:
2. Intelligent Agents
• This behaviour depends on the environment:
• Some environments are more difficult than
others
Zell: Artificial Intelligence (after Russel/Norvig, 3rd Ed.) 2

2
• An agent is anything that can be viewed as
perceiving its environment through sensors and
acting upon that environment through actuators
2.1 Agents and Environments
g p g
• Human agent:
• Sensors: eyes, ears,
and other organs,
• Actuators: hands, legs,
mouth other body partsmouth, other body parts
• Robotic agent:
• Sensors: cameras, sonars, laser range finders;
• Actuators: various motors, gripper
• Percept: an agent’s perceptual inputs at any
time.
Percept sequence: history of percepts
2.1 Agents and Environments
• Percept sequence: history of percepts
• An agent’s actions may depend on its whole
history
• Agent function: mapping of percept sequence to
action: : *f P A
• Agent program: the internal implementation of
the abstract mathematical agent function

3
Example Vacuum Cleaner Miniworld
Simple agent function:
Percept sequence Action
[A, clean] right
• Only two locations: A, B
• Floor clean or dirty
• Percept: [A, clean]
• Actions: right, left, suck
[ , c ea ] g t
[A, dirty] suck
[B, clean] left
[B, dirty] suck
[A, clean], [A, clean]right
[A, clean], [A, dirty] suck
[A, clean], [B, clean]leftg
[A, clean], [B, dirty] suck
[A, dirty], [A, clean] right
…
• A rational agent is “one that does the right thing”.
• …but what is the right thing?
f
2.2 The Concept of Rationality
• Need performance measures.
• Performance measure: objective criterion for
success of an agent’s behaviour
• E.g., performance measure of the vacuum-
cleaner agent could be amount of dirt cleaned
up, amount of time taken, amount of electricity
consumed, cleanliness of floor afterwards, …

4
• Rationality here depends on four things:
1. The performance measure that defines the criterion
of success
2.2 The Concept of Rationality
2. The agent’s prior knowledge of the environment
3. The actions that the agent can perform
4. The agent’s percept sequence to date.
• Rational agent: For each possible percept
sequence, a rational agent should select an
action that is expected to maximize its
performance measure, given the evidence
provided by the percept sequence and whatever
built-in knowledge the agent has.
• Rationality is distinct from omniscience (all-
knowing with infinite knowledge)
• Agents can perform actions in order to modify
Omniscience, Learning, Autonomy
Agents can perform actions in order to modify
future percepts so as to obtain useful information
(information gathering, exploration)
• A rational agent should not only gather
information, but also learn as much as possible
from what it perceives
A t i t if it b h i i• An agent is autonomous if its behavior is
determined by its own experience (with ability to
learn and adapt). Rational agents should be
autonomous.

5
• Specifying the task environment:
• PEAS: Performance measure, Environment,
Actuators Sensors
2.3 The Nature of Environments
Actuators, Sensors
• Example agent: Automated taxi driver:
• Performance measure: Safe, fast, legal, comfortable
trip, maximize profits
• Environment: Roads, other traffic, pedestrians,
customerscustomers
• Actuators: Steering wheel, accelerator, brake, signal,
horn
• Sensors: Cameras, sonar, speedometer, GPS,
odometer, engine sensors, keyboard
• Example Agent: Medical diagnosis system
• Performance measure: Healthy patient, minimize
costs lawsuits
Specifying the Task Environment
costs, lawsuits
• Environment: Patient, hospital, staff
• Actuators: Screen display (questions, tests,
diagnoses, treatments, referrals)
• Sensors: Keyboard (entry of symptoms, findings,
patient's answers)
• Example Agent: Part-picking robot
• Example Agent: Interactive English tutor

6
• Fully observable (vs. partially observable): An
agent's sensors give it access to the complete state
of the environment at each point in time.
Properties of task environments
• Deterministic (vs. stochastic): The next state of the
environment is completely determined by the current
state and the action executed by the agent. (If the
environment is deterministic except for the actions of
other agents, then the environment is strategic)
• Episodic (vs. sequential): The agent's experience is
divided into atomic "episodes" (each episode
consists of the agent perceiving and then performing
a single action), and the choice of action in each
episode depends only on the episode itself.
• Static (vs. dynamic): The environment is
unchanged while an agent is deliberating. (The
environment is semidynamic if the environment
Properties of task environments
environment is semidynamic if the environment
itself does not change with the passage of time
but the agent's performance score does)
• Discrete (vs. continuous): A limited number of
distinct, clearly defined percepts and actions.
• Single agent (vs multiagent): An agent• Single agent (vs. multiagent): An agent
operating by itself in an environment.

7
Task Environment Observ-
able
Deter-
ministic
Episodic Static Discrete Agents
Crossword puzzle Fully Det. Sequ. Static Discrete Single
Chess with a clock Fully Strateg Sequ Semi Discrete Multi
Examples of Task Environments
Chess with a clock Fully Strateg. Sequ. Semi Discrete Multi
Poker Partially Stoch. Sequ. Static Discrete Multi
Backgammon Fully Stoch. Sequ. Static Discrete Multi
Taxi driving Partially Stoch. Sequ. Dynam. Contin. Multi
Medical Diagnosis Partially Stoch. Sequ. Dynam. Contin. Single
Image analysis Fully Det. Sequ. Semi Contin. Single
Part-picking robot Partially Stoch. Sequ. Dynam. Contin. Singlep g y q y g
Refinery controller Partially Stoch. Sequ. Dynam. Contin. Single
English tutor Partially Stoch. Sequ. Dynam. Discrete Multi
• Agent architecture: computing device with
physical sensors and actuators
Agent = architecture + program
2.4 The Structure of Agents
• Agent = architecture + program
• Four basic types in order of increasing
generality:
• Simple reflex agents
• Model-based reflex agents
• Goal-based agents
• Utility-based agents

8
Simple Reflex Agents
• It keeps track of the current state of the world,
using an internal model. It then chooses an
action in the same way as the reflex agent
Model-based reflex agents
action in the same way as the reflex agent.

9
Model-based reflex agents
and action
• A model-based, goal-based agent. It keeps track
of the world state and of a set of goals, and
chooses an action that will (eventually) lead to
Model-based, goal-based agent
chooses an action that will (eventually) lead to
achievement
of its goals.

10
• Goals alone are not enough to achieve high-
quality behaviour in most environments
Some action sequences are better (cheaper
Utility-based Agents
• Some action sequences are better (cheaper,
quicker, more reliable, …) than others.
• A utility function maps a state (or a sequence of
states) onto a real number, which describes the
associated degree of happiness of the agent
• It uses a model of the world, along with a utility
function. Then it chooses the action that leads to
the best expected utility
Utility-based Agents
the best expected utility,
• Expected utility
is computed by
averaging over
all possible
outcome states,
weighted by theweighted by the
probability of the
outcome.

11
• Programming intelligent machines by hand is
extremely time consuming
Better method: program a learning algorithm and
Learning Agents
• Better method: program a learning algorithm and
then teach machines what they should learn
• Example: Supervised learning of artificial neural
networks (ANNs) or support vector machines (SVMs)
• Or better still, let these machines learn by
themselves (under some constraints)themselves (under some constraints)
• Reinforcement learning of artificial neural networks
• Learning is nowadays the big issue in AI!
• A general learning agent, consisting of 4
components:
• Learning
Learning Agents
• Learning
element
• Performance
element
• Critic
• Problem
generator

12
• Learning element: is responsible for making
improvements, uses feedback from the critic on
how the agent is doing and determines how the
Learning Agents
how the agent is doing and determines how the
performance element should be modified
• Performance element: is responsible for
selecting external actions
• Critic: tells the learning element how well the
agent is doing with respect to a performanceagent is doing with respect to a performance
standard (the percept itself does not say so).
• Problem generator: is responsible for suggesting
actions that lead to new information

Ai 02 intelligent_agents(1)

More Related Content

What's hot (20)

Viewers also liked (9)

Similar to Ai 02 intelligent_agents(1) (20)

More from Mohammed Romi (13)

Recently uploaded (20)

Ai 02 intelligent_agents(1)