history of a.i knowledge representation system

Post on 01-Apr-2015

227 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Artificial Intelligence

Contents

History of A.IKnowledge Representation System

The Dark Ages [ The birth of A.I.]: Duration: 1943-56Contributions: First work by Warren McCulloch & Walter

Pitts [ 1943 ]. It was on the central nervous system-a model of neurons of the brain.

Turing, Computing Machinery & intelligence, 1950ENIAC (Electronic Numerical Integrator And Calculator)

by Von Neumann.Shannon, Programming a computer for playing chess,

1950.The Dartmouth College summer workshop on machine

intelligence, Artificial neural networks and automata theory, 1956

Brief History AI (1 of 7)

Rises of A.I: Duration: 1956-late 1960s Contributions:

John McCarthy (inventor of the term Artificial Intelligence) defined the high level language LISP – one of the oldest programming language, which is still in current use.

General Problem Solver (GPS) by Newell & Simon, 1960Human Problem Solving ideas by Newell & Simon, 1972A framework for representing knowledge by Minsky, 1975

Brief History AI (2 of 7)

The Era of unfulfilled promises [ The impact of reality]

Duration: late1960s-early 1970s Contributions:

The Complexity of theorem proving procedures by Cook, 1971

Reducibility Among Combinatorial Problems by Karp 1972

Brief History AI (3 of 7)

The Discovery of expert systems

Duration: 1970s – mid 1980s Contributions:

DENDRAL – the first successful knowledge-based system by Feigenbum, Bachanan & Lederberg.

MYCIN – another expert system by Feigenbum and Shortllife PROSPECTOR – an expert system for mineral exploration

developed by Stanford Research Institute PROLOG – A logic programming language by Colmerauer,

Roussel & Kowalski EMYCIN – Empty MYCIN, a domain-independent version of

MYCIN, developed by Stanford University.

Brief History AI (4 of 7)

The Rebirth of Artificial Neural Networks: 1965 – onward Contributions:

Neural Networks & Physical Systems with Emergent Collective Computational Abilities by Hopfield.

Self-Organized Formation of Topological Correct Feature maps by Kohonen

Parallel Distributed Processing, by Rumelhart & McClelland

The First IEEE International Conference on Neural Networks.

Brief History AI (5 of 7)

Evolutionary computation [Learning by doing]

Duration: early 1970s – onward Contributions:

Adaptation in Natural and Artificial Systems, by Holland

Genetic Programming: On the Programming of the computers by means of Natural Selection by Koza

Evolutionary computation – Towards a new philosophy of machine intelligence by Fogel

Brief History AI (6 of 7)

Computing with Words: late 1980s – onwards Contributions:

Fuzzy sets & Algorithms by ZadehApplication of Fuzzy logic to Approximate Reasoning

using Linguistic Synthesis by MamdaniExpert Systems and Fuzzy Systems, by Negoita.The First IEEE International Conference on Fuzzy

SystemsNeural Networks and Fuzzy Systems by KoskoFuzzy Logic, MATLAB Application Toolbox by the

MathWork, Inc.)

Brief History AI (7 of 7)

Primary objective of A.I: To store knowledge so that programs can process it

and achieve the resemblance of human intelligence.

Knowledge Representation techniquesRule-based Frame-basedSemantic Network, etc.

Knowledge Representation System (1 of 6)

Features: This is the most popular choice for building

knowledge-based systems. Rule is the most commonly used type of

knowledge representation, which can be defined as an IF-THEN structure.

Knowledge Representation system (2 of 6)(Rule-Based)

THEN PART • It is called consequent or conclusion or action.

IF Part• It is called antecedent or premise or condition.

Rules

So, the basic construct is-IF <antecedent>THEN <consequent>

An Example of this constructRULE #1IF the ‘traffic light’ is greenTHEN the action is goRULE #2IF the ‘traffic light’ is redTHEN the action is stop

Knowledge Representation system (3 of 6)(Rule-Based)

A rule can have multiple antecedents joined by the keywords AND, OR or a combination of both.

For example,RULE#3IF ‘age of the customer’ < 18AND ‘cash withdrawal’ > 1000THEN ‘signature of the parent’ is required.

Knowledge Representation system (4 of 6)(Rule-Based)

Object ValueOperator

Each antecedent & consequent has 3 components

RULE#3

IF ‘taxable income’ > 25000THEN ‘Medicare levy’ = ‘taxable income’ * 1.5 / 100

Knowledge Representation system (5 of 6)(Rule-Based)

RelationIF the ‘fuel tank’ is emptyTHEN the ‘car’ is dead

RecommendationIF the ‘season’ is autumnAND the ‘sky’ is cloudyTHEN the ‘advice’ is ‘take an umbrella’

DirectiveIF the ‘car’ is deadAND the ‘fuel tank’ is emptyTHEN the action is ‘refuel the car’

Knowledge Representation system (6 of 6)(What rules can represent?)

• StrategyIF the ‘car’ is dead

THEN the action ‘check the fuel tank’step 1 is complete

IF step 1 is completeAND the ‘fuel tank’ is fullTHEN the action is ‘check the battery’

step 2 is complete.Heuristic

IF the sample is liquidAND the ‘sample pH’ < 6AND the ‘sample smell’ is vinegarTHEN the ‘sample material’ is ‘acetic acid’

Knowledge Representation system (What rules can represent?)

Knowledge-Base

Rule: IF-THEN

Database

FACT

Inference engine

Explanation Facilities

User Interface

User

Basic structure of rule-based expert system

Knowledge Representation system

Disadvantage•Opaque relations between rules•Ineffective search strategy•Inability to learn.

AdvantagesNatural Knowledge representationUniform structureSeparation of knowledge from the inference engineDealing with incomplete and uncertain knowledge. E.g., IF season is AutumnAND sky is cloudyAND wind is lowTHEN forecast is clear {cf 0.1};

forecast is rain {cf 0.9}

Rule-baseExpert system

Advantages and Disadvantages of Rule-based Knowledge Representation

Conflict Resolutions in Rule-Based Systems

What is Conflict?

A situation when two or more actions are found for only one condition

Or, When two or more rules are fired at a time.

Example of Conflict

Rule 1 Rule 2

IF IF

THEN THEN

The Agent has two legs

AND The Agent has two hands

AND The Agent can sleep

The Agent has two legs

AND The Agent has two hands

AND The Agent can sleep

The Agent is a Man The Agent is a Woman

How to Resolve A Conflict?

According to Yoshiaki Shirai and Saburo Tsuji:

They are Three methods to resolve conflict in a rule- based system:

Fire the rule with the Highest priority Fire the rule with the Longest Match Fire the rule with the Data most recently entered

Fire the rule with the Highest priority

How can you define the Highest Priority?

Let’s see how a Robo girl defines priority

A Robo Girl

It’s just an example. I’ll say later about this robot.

Partner Ballroom Dance Robot (OBDR).

The Kosuge-Wang Laboratory in Tohoku Univ. (Division of Mechanical Engineering, Dept. of Bioengineering and Robotics), Nomura Unison Inc., and Torowazo Inc. teamed up to develop this Partner Ballroom Dance Robot(OBDR).

This robot was exhibited at the Prototype Robot Exhibit at EXPO 2005 Aichi Japan.

Agent 1 Will you dance with me?

Robo girl has given the priority 90

Agent 2

Robo girl has given the priority 10 !!

Will you dance with me?

Result is …

Method 1Highest Priority

Rule 1 Rule 2

IF (Priority 90) IF (Priority 10)

THEN THEN

The Agent asks to dance

AND The Agent’s age is below 35

The Agent asks to dance

AND The Agent’s age is above 60

Say ‘Yes’ and Dance Say ‘Sorry Sir!’

Method 2: Longest Match

Rule 1 Rule 2

IF IF

THEN THEN

The Agent is in the front

AND The Agent is a man

The Agent is in the front

AND The Agent is a man

AND The Agent is bowing

Say ‘Hello’ Bow

1. Agent is in the front

2. Agent is a man

3. Agent is bowing

SO…..

Robot is bowing……..

Method 3

Let’s see a photograph of Mr. Bob taken Last Year.

Date of photograph April 24, 2006

Now see a photograph of This Year.

Date of photograph April 24, 2007

Method 3Timestamp

Rule 1 Rule 2

IF IF

THEN THEN

The Question is “Is Mr. Bob bald?”

[08:16PM 4/24/2006]

Say ‘Yes’ Say ‘NO’

The Question is “Is Mr. Bob bald?”

[08:16PM 4/24/2007]

So now on 3/01/2013

Is Mr. Bob bald?

ANSWER IS NO

THANK YOU

top related