AI Chapter II for computer Science students

This is Artificial Inteligence Coures chapter twofor Computer science Student

1. Chapter II

2. Chapter Outline
Introduction to Intelligent Agents
Introduction
Agents and Environments
Acting of Intelligent Agents (Rationality)
Structure of Intelligent Agents
Agent Types
Important Concepts and Terms

3. Introduction
AI has two major roles:
1. Study the intelligent part concerned with humans.
It have two goals
Scientific Goal:
To determine knowledge representation, learning, rule systems search,
and so on
Explain various sorts of realintelligence.

4. Introduction
EngineeringGoal:
TosolverealworldproblemsusingAItechniquessuchasKnowledge
representation,learning,rulesystems,search,andsoon.
Traditionally,computerscientistsandengineershavebeenmore
interestedintheengineeringgoal,while
Psychologists,philosophersandcognitivescientistshavebeenmore
interestedinthescientificgoal.

5. Introduction
2. Designthose actions using computers (Representation).
In order to design intelligent systems, it is important to categorize them
into four categories (Luger and Stubberfield1993), (Russell and Norvig,
2003)
1. Systems that think like humans
2. Systems that think rationally
3. Systems that behave like humans
4. Systems that behave rationally

6. Introduction

7. Introduction
CognitiveScience:ThinkHuman-Like
Requiresamodelforhumancognition.Preciseenoughmodels
allowsimulationbycomputers.
FocusisnotjustonbehaviorandI/O,butlookslikereasoning
process.
Goalisnotjusttoproducehuman-likebehaviorbuttoproducea
sequenceofstepsofthereasoningprocess,similartothesteps
followedbyahumaninsolvingthesametask.

8. Introduction
Laws of thought: Think Rationally
The study of mental faculties through the use of computational models;
that it is, the study of computations that make it possible to perceive
reason and act.
Focus is on inference mechanisms that are probably correct and
guarantee an optimal solution.
Goal is to formalize the reasoning process as a system of logical rules
and procedures of inference.
Develop systems of representation to allow inferences to be like
“Socrates is a man. All men are mortal. Therefore Socrates is mortal”.

9. Introduction
Turing Test: Act Human-Like
The art of creating machines that perform functions requiring intelligence
when performed by people; that it is the study of, how to make computers
do things which, at the moment, people do better.
Focus is on action, and not intelligent behavior centered around the
representation of the world.
Rational agent: Act Rationally
Tries to explain and emulate intelligent behavior in terms of
computational process; that it is concerned with the automation of the
intelligence.
Focus is on systems that act sufficiently if not optimally in all situations.
Goal is to develop systems that are rational and sufficient.

10. Agents and Environments
An agentis an entity that perceives and acts.
Abstractly, an agent is a function from percept histories to
actions:
[f: P* →A]
For any given class of environments and tasks, we seek the agent
(or class of agents) with the best performance.

11. Agents and Environments
Theagentabletosensetheenvironmentwiththehelpofsensors,
theenvironmentpercept/detected/observedbytheagentthrough
sensors.
Thenthesensorssendsthesignalorenergytotheactuators.
Theactuatorsperformstheactionmeanstheyconvertsthissignalor
energyintomotions.
Sotheactionisseenontheenvironment.

12. AI Agent Terminologies
An agent:
Is anything that can perceive/observe/detect its environment through
sensors and acts upon that environment through actuators.
Agents can be three types:
A human agent:
Has sensory organs such as eyes, ears, nose, tongue and skin and has
actuators such as hands, legs, mouth.
A robotic agent:
Has cameras and infrared range finders for the sensors, and various
motors/tires used as actuators.
A software agent:
Has encoded bit strings as its programs and actions (by key strokes).

13. AI Agent Terminologies
Sensor:
Itisadevicewhichperceives/observes/detectsthechangeinthe
environmentandsendstheinformationtotheotherelectronic
devices/actuators/effectors.
Actuators/effectors:
Itiscomponentofmachinethatconvertsenergyintomotion.
Itisresponsibleformovingandcontrollingofthesystem.
Itisthedevicewhichaffectstheenvironment.
Itcanbelegs,wheels,arms,fingersanddisplayscreens.

14. AI Agent Terminologies
Performance Measure of Agent
It is the criteria, which determines how successful an agent is.
Behavior of Agent
It is the action that agent performs after any given sequence of percepts.
Percept
It is agent’s perceptual inputs at a given instance.
Percept Sequence
It is the history of all that an agent has perceived till date.
Agent Function
It is a map from the precept sequence to an action.

15. Agent Rationality
Rationalityisnothingbutstatusofbeingreasonable,
sensible,andhavinggoodsenseofjudgment.
Rationalityisconcernedwithexpectedactionsandresults
dependinguponwhattheagenthasperceived.
Performingactionswiththeaimofobtaininguseful
informationisanimportantpartofrationality.

16. Ideal Rational Agent
It is which has a capable of doing expected actions to maximize its
performance measure, on the basis of:
Its percept sequence
Its built-in knowledge base
They always performs right action, where the right action means the
action that causes the agent to be most successful in the given
percept sequence.
The problem the agent solves is characterized by Performance
Measure, Environment, Actuators, and Sensors(PEAS).

17. Rationality of an agent depends
Theperformancemeasures,whichdeterminethedegree
ofsuccess.
Agent’sPerceptSequencetillnow.
Theagent’spriorknowledgeabouttheenvironment.
Theactionsthattheagentcancarryout.
StructureofIntelligentAgents.

18. Agent’s structure can be viewed as:
Agent = Architecture + Agent Program
Architecture = the machinery that an agent executes on.
Agent Program = an implementation of an agent function.

19. Types of Agents
Agentscanbegroupedintofourclassesbasedontheirdegreeof
perceivedintelligenceandcapability:
oSimpleReflexAgents:CompleteObservationofonlycurrent
state
oModel-BasedReflexAgents:partiallyobservable,dependson
thepercepthistory
oGoal-BasedAgents:selectingtheonewhichreachesagoal
state
oUtility-BasedAgents:itbotherabouthappiness
oLearningAgent:isthetypeofagentthatcanlearnfromits
pastexperiencesorithaslearningcapabilities.

20. Types of Agents
Simple reflex agents
Itworksonlyonthebasisofcurrentperceptionanditdoes
notbotheraboutthepreviousstateinwhichthesystem
was.
Itignoretherestofthepercepthistoryandactonlyonthe
basisofthecurrentpercept.
Percepthistoryisthehistoryofallthatanagenthas
perceivedtilldate.
Itisbasedonthecondition-actionrule
Acondition-actionruleisarulethatmapsastatei.e.,
conditiontoanaction.
Iftheconditionistrue,thenactionistaken,elsenottaken.

21. Types of Agents
Simple reflex agents

22. Types of Agents
Problemswithsimplereflexagentsare:
Theychooseactionsonlybasedonthecurrentpercept.
Theyarerationalonlyifacorrectdecisionismadeonlyon
thebasisofcurrentprecept.
Theirenvironmentiscompletelyobservable.
Verylimitedintelligence.
Noknowledgeofnon-perceptualpartsofstate.Usually
toobigtogenerateandstore.

23. Types of Agents
Model Based Reflex Agent
Itworksbyfindingarulewhoseconditionmatchesthecurrent
situation.
Itcanhandlepartiallyobservableenvironmentsbyuseofmodel
abouttheworld.
Theagenthastokeeptrackofinternalstatewhichisadjustedby
eachperceptandthatdependsonthepercepthistory.
Updatingthestaterequirestheinformationabout:
Howtheworldevolvesin-dependentlyfromtheagent,and
Howtheagentactionsaffectstheworld.

24. Types of Agents
Model Based Reflex Agent

25. Types of Agents
Goal Based Agent
Agents take decision based on how far they are currently from their
goal(description of desirable situations).
Their every action is intended to reduce its distance from goal.
This allows the agent a way to choose among multiple possibilities, selecting the
one which reaches a goal state.
The knowledge that supports its decisions is represented explicitly and can be
modified, which makes these agents more flexible.
Usually require search and planning.
The goal based agent’s behavior can easily be changed.

26. Types of Agents
Goal Based Agent

27. Types of Agents
Utility Based Agents
Theagentswhicharedevelopedhavingtheirendusesas
buildingblocksarecalledutilitybasedagents.
Whentherearemultiplepossiblealternatives,thento
decidewhichoneisbest,utilitybasedagentsareused.
Theychooseactionsbasedonapreference(utility)foreach
state.Sometimesachievingthedesiredgoalisnotenough.
Wemaylookforquicker,safer,cheapertriptoreacha
destination.

28. Types of Agents
UtilityBasedAgents
Agenthappinessshouldbetakenintoconsideration.
Utilitydescribeshow“happy”theagentis.
Becauseoftheuncertaintyintheworld,autilityagentchooses
theactionthatmaximizestheexpectedutility.
Autilityfunctionmapsastateontoarealnumberwhich
describestheassociateddegreeofhappiness.

29. Types of Agents
Utility Based Agents

30. Types of Agents
LearningAgents
AlearningagentinAIisthetypeofagentthatcanlearnfromits
pastexperiencesorithaslearningcapabilities.Itstartstoactwith
basicknowledgeandthenisabletoactandadaptautomatically
throughlearning.Alearningagenthasmainlyfourconceptual
components,whichare:
1.Learningelement:Itisresponsibleformakingimprovementsbylearning
fromtheenvironment.
2.Critic:Thelearningelementtakesfeedbackfromcriticswhichdescribes
howwelltheagentisdoingwithrespecttoafixedperformancestandard.
3.Performanceelement:Itisresponsibleforselectingexternalaction.
4.ProblemGenerator:Thiscomponentisresponsibleforsuggestingactions
thatwillleadtonewandinformativeexperiences.

31. Types of Agents
LearningAgents

32. Interacting with the Environment
Inordertoenableintelligentbehaviour,wewillhavetointeract
withourenvironment.
Properlyintelligentsystemsmaybeexpectedto:
Sensoryinput
Vision,Sound,…
Interactwithhumans
Understand language, recognise speech,
generatetext,speechandgraphics,…
Modifytheenvironment
Robotics

33. Properties of an Environment
Theenvironmenthasmultifoldproperties:
Discrete/Continuous
Iftherearealimitednumberofdistinct,clearly
defined,statesoftheenvironment,theenvironment
isdiscrete(Forexample,chess);otherwiseitis
continuous(Forexample,driving).
CompleteObservable/PartiallyObservable
Ifitispossibletodeterminethecompletestateof
theenvironmentateachtimepointfromthe
perceptsitisobservable;otherwiseitisonly
partiallyobservable.

34. Properties of an Environment
Static/Dynamic
Iftheenvironmentdoesnotchangewhileanagentis
acting,thenitisstatic;otherwiseitisdynamic.
Singleagent/Multipleagents
Theenvironmentmaycontainotheragentswhichmay
beofthesameordifferentkindasthatoftheagent.
Accessible/Inaccessible
Iftheagent’ssensoryapparatuscanhaveaccesstothe
completestateoftheenvironment,thenthe
environmentisaccessibletothatagent.

35. Properties of an Environment
Deterministic/Non-deterministic
Ifthenextstateoftheenvironmentiscompletely
determinedbythecurrentstateandtheactionsofthe
agent,thentheenvironmentisdeterministic;
otherwiseitisnon-deterministic.

36. The Disadvantages of AI
Increased costs
Difficulty with software development -slow and
expensive
Few experienced programmers
Few practical products have reached the market as yet.

37. End of Chapter II
Chapter III Outline
Problem Solving by Searching
Problem Solving Agents
Problem Formulation
Search Strategies
Constraint Satisfaction Search
Games as Search Problems
Next Chapter III