Intelligence Agents
WHAT IS AN AGENT
DEFINITION
Entity that perceives environment using sensors
Acts on environment using actuators
TYPES
HARDWARE
Robots
SOFTWARE
Softbots
EXAMPLES
INPUT (SENOSRS)
Keyboard
Data
OUTPUT (ACTUATORS)
Screens
Files
Networks
PERCEPTS
PERCEPT
what agent senses
PERCEPT SEQUENCE
history of inputs
Decisions depend on percept sequence
Agent cannot use unseen information
AGENT FUNCTION vs PROGRAM
AGENT FUNCTION
Abstract description of behaviour
AGENT PROGRAM
Actual implementation (code)
RUNS ON
Architecture
AGENT ARCHITECTURE
System structure supporting program
PROGRAM
Implements
Agent Function
RATIONAL AGENT
DEFINITION
Does the “right action”
Based on outcomes
USES
PERFOMANCE MEAUSURE
Evaluates environment state
Defines success
Examples
Clean floor
Minimize time & energy
PERCEPT SEQUENCE
PRIOR KNOWLEDGE
AVAILABLE ACTIONS
Read more
https://www.ibm.com/topics/artificial-intelligence
TASK ENVIRONMENT
DEFINITION
Context where agent operates
Defines problem space
P.E.A.S.
P
Performance (success criteria)
E
Environment (where agent works)
A
Actuators (actions)
S
ensors (inputs)
DEFINES
Task Environment
Read more
https://www.geeksforgeeks.org/artificial-intelligence/agents-in-ai/
EXAMPLE: SELF-DRIVING CAR
Performance
safe, fast, legal
Environment
roads, traffic, pedestrians
Actuators
steering, brakes, signals
perform
Actions
Sensors
camera, GPS, speedometer
provide
Percepts
ENVIRONMENT PROPERTIES
Fully / Partially observable
Single / Multi-agent
Deterministic / Stochastic
predictable
Episodic / Sequential
Static / Dynamic
Discrete / Continuous
TYPES OF AGENTS
SIMPLE REFLEX AGENT
condition-action rules
MODEL-BASED AGENT
uses internal world model
GOAL-BASED AGENT
planning & decision-making
UTILITY-BASED AGENT
maximises performance quality
LEARNING AGENTS
DEFINITION
Improve performance over time
COMPONENTS
Learning Element
updates knowledge
Performance Element
selects actions
Critic
gives feedback
Problem Generator
explores new actions
IMPROVES
Rational Behaviour
Intelligence Agents
Intelligence Agents
WHAT IS AN AGENT
DEFINITION
Entity that perceives environment using sensors
Acts on environment using actuators
TYPES
HARDWARE
Robots
SOFTWARE
Softbots
EXAMPLES
INPUT (SENOSRS)
Keyboard
Data
OUTPUT (ACTUATORS)
Screens
Files
Networks
PERCEPTS
PERCEPT
what agent senses
PERCEPT SEQUENCE
history of inputs
Decisions depend on percept sequence
Agent cannot use unseen information
AGENT FUNCTION vs PROGRAM
AGENT FUNCTION
Abstract description of behaviour
AGENT PROGRAM
Actual implementation (code)
RUNS ON
Architecture
AGENT ARCHITECTURE
System structure supporting program
PROGRAM
Implements
Agent Function
RATIONAL AGENT
DEFINITION
Does the “right action”
Based on outcomes
USES
PERFOMANCE MEAUSURE
Evaluates environment state
Defines success
Examples
Clean floor
Minimize time & energy
PERCEPT SEQUENCE
PRIOR KNOWLEDGE
AVAILABLE ACTIONS
Read more
https://www.ibm.com/topics/artificial-intelligence
TASK ENVIRONMENT
DEFINITION
Context where agent operates
Defines problem space
P.E.A.S.
P
Performance (success criteria)
E
Environment (where agent works)
A
Actuators (actions)
S
ensors (inputs)
DEFINES
Task Environment
Read more
https://www.geeksforgeeks.org/artificial-intelligence/agents-in-ai/
EXAMPLE: SELF-DRIVING CAR
Performance
safe, fast, legal
Environment
roads, traffic, pedestrians
Actuators
steering, brakes, signals
perform
Actions
Sensors
camera, GPS, speedometer
provide
Percepts
ENVIRONMENT PROPERTIES
Fully / Partially observable
Single / Multi-agent
Deterministic / Stochastic
predictable
Episodic / Sequential
Static / Dynamic
Discrete / Continuous
TYPES OF AGENTS
SIMPLE REFLEX AGENT
condition-action rules
MODEL-BASED AGENT
uses internal world model
GOAL-BASED AGENT
planning & decision-making
UTILITY-BASED AGENT
maximises performance quality
LEARNING AGENTS
DEFINITION
Improve performance over time
COMPONENTS
Learning Element
updates knowledge
Performance Element
selects actions
Critic
gives feedback
Problem Generator
explores new actions
IMPROVES
Rational Behaviour
Create your own mind map →
Intelligence Agents
Download
Share
Edit
New
Node Notes
Download Mind Map
File format
JPG - Image
SVG - Vector Graphics
PDF - Document
Download
Share Mind Map
Scan the qr code or copy this link to share your mind map.
Copy
Shared by
Sachintha
Terms of Use
-
Privacy Policy
Saving copy…
Sachintha