0% found this document useful (0 votes)
21 views18 pages

AI and Expert System

Very useful

Uploaded by

akilank781
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
21 views18 pages

AI and Expert System

Very useful

Uploaded by

akilank781
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 18

ASSIGNMENT

NAME Akilan.K

REG NO CB22S616355

YEAR |||

STAFF NAME Dr.KARAMCHAND


GANDHI

DEPARTMENT BSC., COMPUTER SCIENCE


KNOWLEDGE REPRESENTATION
❖ Knowledge Representation is a subfield of Artificial Intelligence (AI) that
focuses on how to represent knowledge in a way that computers can
understand and process. It involves creating models and structures to organize
and store knowledge, enabling machines to reason, infer, and make decisions
based on that knowledge.
KNOWLEDGE REPRESENTATION INCLUDES VARIOUS TECHNIQUES, SUCH AS:

✓ Ontologies: Formal Representations Of Knowledge Domains, Defining


Concepts, Relationships, And Rules.
✓ Semantic Networks: Graphical Representations Of Knowledge, Using Nodes
And Edges To Connect Concepts.
✓ Frames: Structured Representations Of Knowledge, Using Slots And Fillers To
Describe Objects And Their Properties.
✓ Rules: If-Then Statements That Encode Knowledge And Enable Reasoning.
✓ Logic: Formal Systems, Like Propositional And Predicate Logic, To Represent
And Reason About Knowledge.

Effective Knowledge Representation enables AI systems to:

✓ Reason And Infer New Information


✓ Make Decisions Based On Available Knowledge
✓ Classify And Categorize Objects And Concepts
✓ Understand Natural Language
✓ Integrate And Reconcile Multiple Sources Of Knowledge

Knowledge Representation is crucial in various AI applications,


including:
✓ Expert Systems
✓ . Natural Language Processing (NLP)
✓ Computer Vision
✓ Robotics
✓ Decision Support Systems

KNOWLEDGE REPRESENTATION OF
ISSUES
KNOWLEDGE ACQUISITION :

❖ Gathering and encoding knowledge from experts or


data sources can be time-consuming and difficult.
ONTOLOGY ALIGNMENT :

➢ Integrating Multiple Ontologies Or Knowledge Models


Can Be Challenging Due To Differences In
Representation, Terminology, Or Structure.
KNOWLEDGE UPDATES :

➢ Keeping Knowledge Representations Up-To-Date And


Consistent With Changing Information Or Domains Can
Be Problematic.
SCALABILITY :

➢ KR Systems Can Become Complex And Difficult To


Manage As The Amount Of Knowledge Grows.
REASONING EFFICIENCY :

➢ Inference And Reasoning Processes Can Be


Computationally Expensive, Leading To
Performance Issues.
AMBIGUITY AND UNCERTAINTY :

➢Handling Ambiguous Or Uncertain


Knowledge Can Be Difficult, Especially In
Complex Domains.
KNOWLEDGE SHARING:

➢ Representing Knowledge In A Way That’s


Shareable And Reusable Across Different Systems
Or Applications Can Be Challenging.
DOMAIN LIMITATIONS :

➢KR systems may not perform well when


applied to domains with diverse or nuanced
knowledge.
LACK OF COMMON SENSE :

➢KR systems may not possess the same level


of common sense or real-world experience as
humans.
Evaluation Metrics :

➢ Assessing The Quality And


Effectiveness Of KR Systems Can Be
Difficult Due To The Lack Of Standard
Evaluation Metrics.
USING PREDICTED LOGIC
❖ Predictive Logic Is A Subfield Of Logic That Deals With Making
Predictions Or Probabilistic Statements About Future Events
Or Outcomes. It Combines Elements Of Probability Theory,
Statistics, And Logic To Reason About Uncertain Events.

➢ Using predictive logic in Knowledge Representation can help


address some of the challenges I mentioned earlier, such as:

1) Handling uncertainty and ambiguity


2) Making probabilistic predictions
3) Reasoning about future events
4) Updating knowledge based on new evidence

PREDICTIVE LOGIC CAN BE APPLIED IN VARIOUS WAYS, INCLUDING:

1) Bayesian networks
2) Markov logic networks
3) Probabilistic ontologies
4) Temporal logic

THESE APPROACHES ENABLE KR SYSTEMS TO:


1) Represent uncertain knowledge
2) Make probabilistic predictions
1) Reason About Temporal Relationships
2) Update Knowledge Based On New Evidence

PREDICTIVE LOGIC HAS APPLICATIONS IN AREAS LIKE:

1) Predictive Analytics
2) Decision Support Systems
3) Expert Systems
4) Natural Language Processing (NLP)

KEY CONCEPTS :

o PROBABILITY : A Measure Of The Likelihood Of An Event Occurring.


o PREDICTION : A Statement About A Future Event Or Outcome.
o INFERENCE : The Process Of Drawing Conclusions Based On Evidence.
o UNCERTAINTY : The State Of Having Limited Or Imperfect Knowledge.

TECHNIQUES :

1. BAYESIAN INFERENCE : Updating Probabilities Based On New Evidence.

2. MARKOV CHAINS : Modeling Sequences Of Events.

3. DECISION TREES : Representing Decisions And Their Outcomes.

4. PROBABILISTIC GRAPHICAL MODELS : Visualizing Relationships


Between Variables.
APPLICATIONS :

1. PREDICTIVE MAINTENANCE : Forecasting equipment failures.

2. MEDICAL DIAGNOSIS : Predicting disease likelihood.

3. FINANCIAL FORECASTING : Predicting market trends.

4. NATURAL LANGUAGE PROCESSING : Predicting word or sentence


likelihood.

BENEFITS :

1. IMPROVED DECISION-MAKING : By Considering Uncertainty And


Probability.

2. ENHANCED RISK ASSESSMENT : By Predicting Potential Outcomes.

3. INCREASED EFFICIENCY : By Identifying Likely Events Or Outcomes.

4. BETTER RESOURCE ALLOCATION : By Prioritizing Likely Scenarios.

CHALLENGES :

1. DATA QUALITY : Ensuring Accurate And Reliable Data.

2. MODEL COMPLEXITY : Balancing Simplicity And Accuracy.

3. INTERPRETABILITY : Understanding And Explaining Predictions.

4. UNCERTAINTY REPRESENTATION : Effectively Capturing Uncertainty.


REPRESENTING KNOWLEDGE
❖ Representing Knowledge Involves Capturing And Organizing
Information In A Way That Computers Can Understand And
Process. Here Are Some Key Aspects Of Representing
Knowledge:

1. ONTOLOGIES : Define
concepts, relationships, and rules to
represent a domain’s knowledge.

2. SEMANTIC NETWORKS : Graphical


representations of
knowledge, using nodes and edges to connect concepts.

3. FRAMES : Structured representations of knowledge, using


slots and fillers to describe objects and their properties.

4. RULES : If-then
statements that encode knowledge and
enable reasoning.

5. LOGIC : Formalsystems, like propositional and predicate


logic, to represent and reason about knowledge.

6. CONCEPT MAPS : Visual representations of knowledge, highlighting


relationships between concepts.
7. KNOWLEDGE GRAPHS : Graphical
representations of
knowledge, emphasizing relationships between entities.

Effective knowledge representation enables:

1. REASONING : Drawing Conclusions From Represented Knowledge.

2. INFERENCE : Making Predictions Or Deductions Based On Knowledge.

3. DECISION SUPPORT : Providing Insights For Informed Decision-Making.

4. NATURAL LANGUAGE UNDERSTANDING : Enabling COMPUTERS TO COMPREHEND


HUMAN LANGUAGE.

5. KNOWLEDGE SHARING : Facilitating The Exchange Of Knowledge Between Systems Or


Humans.

USING RULES

❖ Using rules in knowledge representation involves defining if-then statements


that encode knowledge and enable reasoning. Rules can be used to:

1. Represent expertise : Capture expert knowledge and decision-making processes.

2. REASON ABOUT DATA : Make inferences and draw conclusions from data.

3. Make decisions : Provide decision support and recommendations.

4. Classify objects : Categorize objects based on their properties and characteristics.


5. Detect patterns : Identify patterns and relationships in data.

TYPES OF RULES:

1. _Production rules : If-then statements that assert a conclusion based on conditions.

2. _Inference rules_: Rules that enable reasoning and drawing conclusions.

3. _Terminological rules_: Rules that define relationships between concepts and terms.

4. _Assertional rules_: Rules that assert facts and statements about the world.

RULE-BASED SYSTEMS CAN BE USED IN VARIOUS APPLICATIONS,


INCLUDING:

1. Expert systems : Mimic expert decision-making processes.

2. Decision support systems : Provide recommendations and guidance.

3. Natural language processing : Understand and generate human language.

4. Image recognition : Classify and recognize objects in images.

5. Robotics : Control and navigate robots based on rules and reasoning.

Logic & slot and filler

❖ Logic and slot-and-filler are two


related concepts in knowledge
representation:
LOGIC :

▪ A formal system for reasoning and inference


▪ Used to represent knowledge and make decisions

CONSISTS OF :

▪ Propositional logic (statements and their combinations)


▪ Predicate logic (statements with variables and quantifiers)

SLOT-AND-FILLER

▪ A knowledge representation technique


▪ Used to describe objects and their properties
▪ Consists of:
▪ Slots (attributes or properties)
▪ Fillers (values or instances of those properties)

Example:

▪ Object: Person
▪ Slots: Name, Age, Occupation
▪ Fillers: John, 30, Doctor

In this example, “Person” is the object, “Name”, “Age”, and “Occupation” are the slots, and
“John”, “30”, and “Doctor” are the fillers.

Slot-and-filler is used in various knowledge representation formalisms, such as:

▪ Frames (a structured representation of knowledge)


▪ Semantic networks (a graphical representation of knowledge)
▪ Ontologies (a formal representation of knowledge domains)

The combination of logic and slot-and-filler enables reasoning and inference about objects
and their properties, allowing for:

▪ Classification
▪ Inheritance
▪ Defaults
▪ Reasoning about relationships

GAME PLAYING
❖ Merging Knowledge Representation With Game Playing Can
Lead To Some Exciting Applications:

1. GAME AI : Creating intelligent agents that can play games like chess, poker, or video
games.

2. DECISION-MAKING : Using knowledge representation to inform decision-making in


games.

3. STRATEGY DEVELOPMENT : Representing game strategies and tactics to


improve gameplay.

4. PLAYER MODELING : Modeling player behavior and preferences to personalize


game experiences.

5. GAME GENERATION : Generating new games or levels using knowledge representation


techniques.

SOME KNOWLEDGE REPRESENTATION TECHNIQUES USED IN GAME


PLAYING INCLUDE:
✓ Decision trees
✓ Game trees
✓ Bayesian networks
✓ Markov decision processes
✓ Ontologies

THESE TECHNIQUES CAN BE APPLIED TO VARIOUS GAMES, SUCH AS:

o Board games_ (chess, Go, etc.)


o Card games_ (poker, blackjack, etc.)
o Video games_ (platformers, RPGs, etc.)
o Puzzle games_ (Sudoku, crosswords, etc.)

BY COMBINING KNOWLEDGE REPRESENTATION WITH GAME


PLAYING, WE CAN CREATE:

o Smarter game AI
o More realistic player behavior
o Improved game difficulty adjustment
o Enhanced game analysis and commentary
o New game genres and mechanics

MINIMAL SEARCH
❖ Minimal search refers to the process of finding the shortest or
most efficient path to a solution in a problem space. It’s a
fundamental concept in artificial intelligence, computer
science, and optimization.
Types of minimal search:

1. Uninformed search : No additional information is available, and the search explores the
problem space blindly. (e.g., Breadth-First Search, Depth-First Search)

2. Informed search : Heuristics or additional information guide the search towards the
solution. (e.g., Greedy Search, A* Search)

MINIMAL SEARCH ALGORITHMS:

1. Breadth-First Search (BFS) : Explores all nodes at the current depth before moving
to the next level.

2. Depth-First Search (DFS) : Explores as far as possible along each branch before
backtracking.

3.Dijkstra’s algorithm : Finds the shortest path in a graph with non-negative edge
weights.

4. A algorithm : Combines BFS and DFS with an admissible heuristic function to find
the shortest path.

5. Greedy algorithm: Makes locally optimal choices to find a global optimum.

MINIMAL SEARCH APPLICATIONS :

1. Pathfinding : Finding the shortest path in video games, GPS navigation, or network
routing.

2. Problem-solving : Solving puzzles, such as Sudoku, crosswords, or logic grids.

3. Optimization : Finding the minimum or maximum of a function in fields like operations


research or machine learning.

4. Robotics : Planning motion paths for robots to avoid obstacles.


ALPHA BETA CUTTOFFS

➢ Alpha-Beta Cutoffs Are A Optimization Technique Used In The


Alpha-Beta Pruning Algorithm, Which Is A Popular Algorithm For
Minimizing The Number Of Nodes To Be Evaluated In A Search
Tree.

Alpha-Beta Pruning Works By Maintaining Two Values:

▪ Alpha (α): the best possible score for the maximizing player (usually
white)
▪ Beta (β): the best possible score for the minimizing player (usually
black)

➢ The Algorithm Prunes Branches That Will Not Affect The Final Decision, Reducing
The Search Space.

CUTOFFS OCCUR WHEN :

▪ Alpha ≥ Beta (beta cutoff): the maximizing player will not choose this
branch, so prune it
▪ Alpha ≤ Beta (alpha cutoff): the minimizing player will not choose this
branch, so prune it

TYPES OF CUTOFFS :
▪ Alpha cutoff: the maximizing player’s best score is already better than
the minimizing player’s best score
▪ Beta cutoff: the minimizing player’s best score is already better than
the maximizing player’s best score

BENEFITS OF ALPHA-BETA CUTOFFS :

▪ Reduce The Number Of Nodes To Be Evaluated


▪ Speed Up The Search Process
▪ Improve The Efficiency Of The Algorithm

Common applications:

▪ Game playing (e.g., chess, checkers)


▪ Decision-making
▪ Problem-solving
▪ AI and machine learning

GOAL STACK PLANNING

❖ GOAL STACK PLANNING (GSP) IS A PLANNING TECHNIQUE USED IN ARTIFICIAL


INTELLIGENCE TO SOLVE COMPLEX PROBLEMS BY BREAKING THEM DOWN INTO
SMALLER, MANAGEABLE SUBGOALS. IT’S PARTICULARLY USEFUL FOR PLANNING
IN DYNAMIC ENVIRONMENTS.

KEY COMPONENTS OF GSP :


1. GOAL STACK : A Hierarchical Structure Representing The
Problem’s Goals And Subgoals.

Breaking Down High-Level Goals


2. GOAL DECOMPOSITION :

Into Smaller, More Achievable Subgoals.


3. SUBGOAL GRAPH : A graph representing the relationships
between subgoals.
4. PLANNING : Finding a sequence of actions to achieve the goals.

GSP PROCESS:

1. Goal Identification : Identify the high-level goal.

2. Goal Decomposition : Break down the goal into subgoals.

3. Subgoal Graph Construction : Create a graph representing subgoal relationships.

4. Planning : Find a plan to achieve the goals.

5. Execution : Execute the plan.

6. Monitoring : Monitor progress and adjust the plan as needed.

GSP BENEFITS:

1. Flexibility : Handles changing environments and goals.

2. Efficiency : Reduces planning time by focusing on relevant subgoals.

3. Scalability : Solves complex problems by breaking them down into manageable pieces.

APPLICATIONS OF GSP:
1. Robotics : Planning robot actions in dynamic environments.

2. Game Playing : Planning game strategies and moves.

3. Autonomous Systems : Planning for autonomous vehicles and drones.

4. Decision Support Systems : Supporting decision-making in complex domains.

You might also like