Artificial Intelligence

24.04.2025
─

Prof. Saba
Govt. Graduate College of Commerce, Kamoke
 1

Fuzzy System
Here’s a clear and concise explanation of the Fuzzy System, its
theory, and how fuzzy inference is used in Artificial Intelligence
(AI):

---

1. Fuzzy System Theory

Fuzzy system theory deals with reasoning that is approximate

rather than fixed and exact. It’s based on fuzzy logic, which
allows for degrees of truth (i.e., values between 0 and 1), unlike
classical Boolean logic which only allows true (1) or false (0).

Key Concepts:

Fuzzy Set: A set where elements have degrees of membership.

 2

E.g., the term “hot” temperature: 30°C might be 0.7 “hot”, while
40°C is 1.0 “hot”.

Membership Function: Defines how each input is mapped to a

value between 0 and 1.

Linguistic Variables: Variables described using words (like

“low”, “medium”, “high”).

---

2. Fuzzy Inference in Artificial Intelligence

Fuzzy inference is the process of formulating the mapping from

a given input to an output using fuzzy logic. It is widely used in
AI to deal with uncertainty and approximate reasoning,
especially in expert systems, decision-making, and control
systems.
 3

Steps in Fuzzy Inference:

1. Fuzzification: Convert crisp inputs into fuzzy values.

2. Rule Evaluation (Inference): Apply fuzzy rules from a rule

base.

E.g., IF temperature is high AND humidity is low THEN fan

speed is high

3. Aggregation: Combine the results of all rules.

4. Defuzzification: Convert the fuzzy output into a crisp value.

Applications in AI:

●​ Smart appliances (e.g., air conditioners)

●​ Autonomous vehicles
 4

●​ Medical diagnosis systems

●​ Natural language processing

---

Example: Fuzzy Inference System for Controlling a Fan

Goal: Control the fan speed based on temperature and humidity

using fuzzy logic.

---

1. Inputs and Output

Inputs (crisp values):

Temperature: 30°C
Humidity: 40%
 5

Output:
Fan Speed

---

2. Fuzzification

Convert inputs into fuzzy values using membership functions:

Temperature Membership Functions:

Low: 0°C–20°C

Medium: 15°C–30°C

High: 25°C–40°C

At 30°C:

Medium: 0.4
 6

High: 0.6

Humidity Membership Functions:

Low: 0%–50%

High: 40%–100%

At 40%:

Low: 0.8

High: 0.2

---

3. Fuzzy Rules
 7

Sample fuzzy rules:

1. IF Temp is High AND Humidity is Low THEN Fan Speed is

High

2. IF Temp is Medium AND Humidity is Low THEN Fan Speed

is Medium

3. IF Temp is Low AND Humidity is High THEN Fan Speed is

Low

---

4. Inference Engine

Apply the rules using min operator (AND = min):

 8

Rule 1: min(High Temp 0.6, Low Humidity 0.8) = 0.6 → Fan

Speed = High (0.6)

Rule 2: min(Medium Temp 0.4, Low Humidity 0.8) = 0.4 →

Fan Speed = Medium (0.4)

---

5. Aggregation

Combine the outputs from all active rules:

Fan Speed has:
High: 0.6
Medium: 0.4

---
 9

6. Defuzzification
Convert fuzzy output to a crisp value using centroid method:

Final fan speed = Weighted average of the output values.

---

Diagram

I’ll create a visual representation of this fuzzy inference system.

One moment…

It looks like you might be referring to "mining frequent patterns"

— a key concept in data mining. Let me explain it clearly:

---
 10

Frequent Pattern Mining

Frequent pattern mining is the process of finding patterns (like

sets of items, subsequences, or substructures) that appear
frequently in a dataset.

Common in:

●​ Market Basket Analysis

●​ Web Usage Mining
●​ Bioinformatics
●​ Intrusion Detection

---

Key Concepts:
 11

Itemset: A collection of one or more items (e.g., {milk, bread})

Support: How often an itemset appears in the dataset.

Minimum Support Threshold: Only itemsets appearing more

frequently than this threshold are considered frequent.

---

Example: Market Basket Analysis

Suppose we have a set of transactions like:

 12

If the support threshold is 50%, the algorithm will find patterns

like:

{Milk} → appears in 3 out of 4 transactions → 75% → frequent

{Bread, Milk} → 2 out of 4 → 50% → frequent

 13

---

Popular Algorithms:

Apriori: Classic algorithm, generates candidate sets and filters

them by support.

FP-Growth (Frequent Pattern Growth): More efficient, uses a

tree-based structure called FP-Tree.

---
 14

Association and correlation:

1. Association

Association refers to rules that imply relationships between

items in large datasets.

Used in:
●​ Market basket analysis (e.g., Amazon, Walmart)
●​ Recommendation systems
●​ Retail analytics

Example:

If many customers who buy milk also buy bread, we can form a
rule:

"Milk → Bread"
 15

This doesn’t mean milk causes bread to be bought — it just

shows they occur together frequently.

Key Metrics:

Support: How often the items appear together.

Confidence: How often the rule is true.
Lift: How much more likely bread is bought if milk is bought,
compared to chance.

---

2. Correlation

Correlation measures the strength and direction of a relationship

between two numerical variables.

Used in:
 16

●​ Statistical analysis
●​ Predictive modeling
●​ Data science

Example:

If the number of hours studied increases, and exam scores also

increase, they are positively correlated.

Correlation Coefficient (r):

r = +1 → perfect positive correlation

r = 0 → no correlation

r = –1 → perfect negative correlation

Diagram:
visual comparison or a real dataset example to see both in action
17
 18

What is Evolutionary Computing?

Evolutionary Computing is a subfield of artificial intelligence

(AI) that uses natural selection and genetics-inspired techniques
to solve complex problems.

It mimics biological evolution — survival of the fittest — to

evolve better solutions over time.

---

Key Features:

●​ Population of potential solutions

●​ Fitness evaluation
●​ Selection based on performance

Genetic operations: mutation, crossover (recombination), and

selection
 19

---

Main Techniques in Evolutionary Computing:

1. Genetic Algorithms (GA):

Most popular.

Uses binary strings to represent solutions.

2. Genetic Programming (GP):

Evolve computer programs or expressions.

3. Evolution Strategies (ES):

Focused on real-valued optimization.

4. Differential Evolution (DE):
 20

Good for numerical problems.

5. Evolutionary Programming (EP):

Originally for evolving finite state machines.

---

General Steps in Evolutionary Algorithm:

1. Initialization: Create a random population.

2. Evaluation: Measure fitness of each individual.
3. Selection: Choose better solutions.
4. Crossover & Mutation: Create new individuals.
5. Replacement: Form the next generation.
6. Termination: Repeat until stopping condition is met.
 21

---

Applications:
●​ Optimizing design parameters
●​ Machine learning hyperparameter tuning
●​ Game playing AI
●​ Robotics
●​ Scheduling and planning

---

What is a Genetic Algorithm (GA)?

A Genetic Algorithm is an optimization technique inspired by

natural evolution. It finds good solutions to problems by
simulating the process of natural selection.
 22

---

Key Concepts:

Chromosome (Individual): A possible solution (usually encoded

as a string of bits, numbers, or symbols).

Population: A group of chromosomes.

Fitness Function: Measures how good a solution is.

Selection: Chooses the fittest individuals for reproduction.

Crossover (Recombination): Combines parts of two

chromosomes to create a new one.

Mutation: Randomly changes part of a chromosome to maintain

diversity.

Generation: One cycle of selection, crossover, and mutation.

 23

---

How GA Works (Step-by-Step):

1. Initialization:
Generate a random population of solutions.

2. Evaluation:
Use the fitness function to evaluate each solution.

3. Selection:
Select the fittest individuals for reproduction (e.g., roulette
wheel or tournament selection).

4. Crossover:
Combine parts of selected individuals to create offspring.
 24

5. Mutation:
Randomly change parts of offspring to introduce variation.

6. Replacement:
Replace the old population with the new one.

7. Termination:
Stop if a solution is good enough or after a certain number of
generations.

---

Example:

Problem: Maximize the number of 1s in a binary string of

length 8.
Chromosome: 10101011
Fitness: Number of 1s = 6
 25

---

Applications:

●​ Engineering design
●​ Scheduling problems
●​ Game strategies
●​ Feature selection in machine learning
●​ Optimization in AI systems

---

Applications of Genetic Algorithms

Genetic Algorithms are powerful for solving optimization

and search problems where traditional methods struggle.
 26

Common Applications:

1. Function Optimization:

Find the maximum/minimum of complex mathematical

functions.

2. Scheduling Problems:
Optimize task or job schedules (e.g., timetables, airline
schedules).

3. Route Planning:
Solve problems like the Traveling Salesman Problem
(TSP).

4. Game Playing & AI:

Evolve strategies for bots or characters.
 27

5. Machine Learning:

Feature selection, hyperparameter tuning.

6. Robotics:

Path planning and control system design.

7. Engineering Design:

Structural optimization, antenna design.

---

Example: Maximize the Number of 1s in a Binary String

 28

Problem:

Find a binary string of length 6 that has the maximum

number of 1s.

Chromosome:

Each solution is a 6-bit binary string.

E.g., 101011, 110110, 000111

Fitness Function:

Number of 1s in the string.

E.g., 101011 → fitness = 4

Steps:

1. Initial Population: Random strings like:

100101, 111000, 001111

 29

2. Evaluate Fitness:

Count number of 1s in each.

3. Selection:

Pick the best (most 1s) for reproduction.

4. Crossover:

Combine parts of parents:

Parent 1: 101100
Parent 2: 110011
Offspring: 101011

5. Mutation:

Randomly flip a bit: `101

 30

In Genetic Algorithms (GA), genetic operators are the core

mechanisms that mimic natural evolution. They are used to
generate new solutions from existing ones.

---

Main Genetic Operators

1. Selection
Purpose: Choose the best individuals (parents) for
reproduction based on fitness.

Common methods:
 31

Roulette Wheel Selection: Probability of selection is

proportional to fitness.

Tournament Selection: Randomly select a group, and

choose the best from them.

Rank Selection: Rank individuals and select based on their

position.

---

2. Crossover (Recombination)

Purpose: Combine genetic material from two parents to

produce offspring.

Types:
 32

Single-Point Crossover: A point is selected, and the tail

parts are swapped.

Parent1: 110|101

Parent2: 011|010

Offspring: 110010

Two-Point Crossover: Two points are chosen, and the

middle part is swapped.

Uniform Crossover: Each gene is randomly selected from

one of the parents.
 33

---

3. Mutation

Purpose: Introduce diversity and prevent premature

convergence.

How it works: Randomly flips bits or changes values.

E.g., 101011 → 100011 (flipped 2nd bit)

---

4. Replacement

Purpose: Decide which individuals survive to the next

generation.
 34

Strategies:

●​ Replace the worst individuals.

●​ Keep some best ones (elitism).
●​ Replace randomly.

---

5. Elitism (optional)

Purpose: Ensure the best solutions are carried forward.

How: Keep top 1 or more individuals unchanged in each
generation.
 35

---