Department of CSE(AI&ML) II Year B.

Tech-I Semester

1. Course Objectives
1. The student should be made to study the concepts of Artificial Intelligence.
2. The student should be made to learn the methods of solving problems using Artificial
Intelligence.
3. The student should be made to introduce the concepts of Expert Systems.
4. To understand the applications of AI, namely game playing, theorem proving, and
machine learning.
5. To learn different knowledge representation techniques

2. Pre-requisite
1. Knowledge in Computer Programming.
2. A course on “Mathematical Foundations of Computer Science”.
3. Background in linear algebra, data structures and algorithms, and probability.

Compiled By: P.Rajyalakshmi 1

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

3. Syllabus
L T P C
(23A31301T) ARTIFICIAL INTELLIGENCE 3 0 0 3

UNIT - I

Introduction: AI problems, foundation of AI and history of AI intelligent agents: Agents and

Environments, the concept of rationality, the nature of environments, structure of agents,
problem solving agents, problem formulation.

UNIT - II

Searching- Searching for solutions, uniformed search strategies – Breadth first search, depth
first Search. Search with partial information (Heuristic search) Hill climbing, A*, AO*
Algorithms, Problem reduction, Game Playing-Adversial search, Games, mini-max
algorithm, optimal decisions in multiplayer games, Problem in Game playing, Alpha-Beta
pruning, Evaluation functions.

UNIT - III

Representation of Knowledge: Knowledge representation issues, predicate logic- logic

programming, semantic nets- frames and inheritance, constraint propagation, representing
knowledge using rules, rules-based deduction systems. Reasoning under uncertainty, review
of probability, Bayes’ probabilistic interferences and dempster Shafer theory.

UNIT - IV

Logic concepts: First order logic. Inference in first order logic, propositional vs. first order
inference, unification & lifts forward chaining, Backward chaining, Resolution, learning from
observation Inductive learning, Decision trees, Explanation based learning, Statistical
Learning methods, Reinforcement Learning.

UNIT - V

Expert Systems: Architecture of expert systems, Roles of expert systems – Knowledge

Acquisition Meta knowledge Heuristics. Typical expert systems – MYCIN, DART, XCON:
Expert systems shells.

Compiled By: P.Rajyalakshmi 2

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Textbooks:

1. S. Russel and P. Norvig, “Artificial Intelligence – A Modern Approach”, Second

Edition, Pearson Education.
2. Kevin Night and Elaine Rich, Nair B., “Artificial Intelligence (SIE)”, Mc Graw Hill

Reference Books:

1. David Poole, Alan Mackworth, Randy Goebel,” Computational Intelligence: a logical

approach”, Oxford University Press.
2. G. Luger, “Artificial Intelligence: Structures and Strategies for complex problem
solving”, Fourth Edition, Pearson Education.
3. J. Nilsson, “Artificial Intelligence: A new Synthesis”, Elsevier Publishers.
4. Artificial Intelligence, Saroj Kaushik, CENGAGE Learning.

Online Learning Resources:

1. https://ai.google/
2. https://swayam.gov.in/nd1_noc19_me71/preview

Compiled By: P.Rajyalakshmi 3

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

4. Course Outcomes

Cognitive
Course Course Outcome /affective Course Outcome
Code Statement Level of the
course
outcome
Explain the foundations of
23A31301T AI and various Intelligent Apply L3 CO1
Agents

Apply search strategies in

23A31301T problem solving and game Apply L3 CO2
playing

Apply problem solving

23A31301T Apply L4 CO3
strategies with knowledge
representation techniques for
applications of AI

Explain logical agents and

23A31301T first-order logic Apply L3 CO4

Describe the basics of

23A31301T Learning and Expert Apply L4 CO5
Systems

Compiled By: P.Rajyalakshmi 4

S.V.P.C.E.T
 Department of CSE(AI&ML) II Year B. Tech-I Semester

5. LECTURE PLAN
UNIT I

S.No. Topic No. of Proposed Actual Pertaining Taxonomy Mode of

Periods lecture lecture CO(s) Level Delivery

Period Period
1 Introduction- AI
Problems 1 CO1 L2 MD1,
MD5
2 Foundation of
AI, History of AI 1 CO1 L1 MD1,
MD5

3 Intelligent
Agents and 1 CO1 L2 MD1,
Environments MD5
4 The Concept of
Rationality 1 CO1 L2 MD1,
MD5
5 Nature of
environments, 1 CO1 L2 MD1,
Structure of MD5
Agents

6 Problem solving 1
agents, CO1 L2 MD1,
Problem MD5
formulation
7 Problem solving 1
agents-Example CO1 L3 MD1,
Problems MD5

Compiled By: P.Rajyalakshmi 5

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

UNIT - I

Introduction: AI problems, foundation of AI and history of AI intelligent agents: Agents

and Environments, the concept of rationality, the nature of environments, structure of
agents, problem solving agents, problem formulation.

1. Introduction:
Artificial Intelligence (AI) refers to the “development of computer systems for
performing tasks that require human intelligence”. The goal of AI is to create
machines to carry out diverse tasks for identifying patterns and making decisions based
on the collected information. This can be achieved through AI techniques (machine
learning and neural networks to natural language processing and robotics) which
enable machines to learn, reason, and perform tasks previously reserved for human intelligence.
The field of AI is unique, sharing border with disciplines like Mathematics, Computer
Science, Philosophy, Psychology, Biology, Cognitive Science and many others.

 Although there is no clear definition of AI or even Intelligence, it can be described

as an attempt “to build machines that like humans can think and act, able to learn and
use knowledge to solve problems” on their own.

 Artificial Intelligence (AI), a term coined by emeritus Stanford Professor John

McCarthy in 1956, was defined by him as “the science and engineering of making
intelligent machines”

Compiled By: P.Rajyalakshmi 6

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Definition: Artificial intelligence, commonly known as AI, is a branch of

computer science that aims to develop systems capable of performing tasks that
normally require human intelligence.
1.1 Organization of AI:

To design intelligent systems, it is important to categorize them into four categories

(Luger and Stubberfield 1993), (Russell and Norvig, 2003) by definitions:

The four approaches in more detail are as follows:

1. Systems that think like humans (The cognitive modelling approach): The idea
behind this approach is to determine whether the computer thinks like a human. It is an
area of psychology/cognitive science. One must know functioning of brain and its
mechanism for processing information. Neural network is a computing model for
processing information similar to brain.
2. Systems that act like humans (The Turing Test approach): This approach was
designed by Alan Turing in 1950.
• The ideology behind this approach is that “a computer passes the test if a human
interrogator, after asking some written questions, cannot identify whether the written
responses come from a human or from a computer”.

Compiled By: P.Rajyalakshmi 7

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

• This test is used to evaluate a computer acting like humans.

The features required for a machine to pass the Turing test:

 Natural language processing to enable it to communicate successfully in English (or
some other human language) with the interrogator.
 Knowledge representation to store information provided before or during the
interrogation.
 Automated reasoning to use the stored information to answer questions and to draw
new conclusions.
 Machine learning to adapt to new circumstances and to detect and infer patterns.

Total Turing Test includes a video signal so that the interrogator can test the
subject’s perceptual abilities, as well as the opportunity for the interrogator to pass
physical objects ―” through the hatch”. To pass the total Turing Test, the computer
will need:

 Computer Vision to perceive objects in the environment i.e. to recognize the

interrogator’s actions and other objects during a test.
 Robotics to fulfill mechanical tasks.
3. Systems that think rationally (The laws of thought approach): Such systems rely
on logic rather than human to measure correctness. For thinking rationally or
logically, logic formulas and theories are used for synthesizing outcomes.
For example: “John is a human, and all humans are mortal”
then one can conclude logically that, “John is mortal”

Compiled By: P.Rajyalakshmi 8

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

4. Systems that act rationally (The rational agent approach): It tries to define AI,
with the concept of so-called rational agents. An agent is just something that acts and
a rational agent is one that acts to achieve the best outcome when there is uncertainty.
Rational behavior means doing right thing.

1.2 AI Problems:

Problems in Artificial Intelligence (AI) come in different forms, each with its own
set of challenges and potential for innovation. From image recognition to natural
language processing, AI problems can be classified into three types of tasks:
1. Mundane Tasks: These tasks are done routinely by people and some other
animals. (For e.g. Perception, robotics, natural language, Vision, Speech)
o Recognizing people, objects.
o Communicating (through natural language).
o Navigating around obstacles on the streets.

2. Formal Tasks: Tasks which require logic and constraints to function. (For e.g.
Mathematics-logic, calculus, algebra Theorem proving, Games chess,
checkers, verification,).

Compiled By: P.Rajyalakshmi 9

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

3. Expert tasks: These tasks cannot be done by all people and can only be
performed by skilled specialists. (For e.g. financial analysis, medical
diagnostics, engineering, scientific analysis, consulting)
 Much of the early work in the field of AI was focused on formal tasks, such as Game
Playing and Theorem proving. Initially computers could perform well at those tasks,
but no computer is fast enough to overcome the combinational explosion
generated by most problems.
 Another early foray into AI focused on common-sense reasoning. Again, no attempt
was made to create a program with a large amount of knowledge.
 This is the reason why AI work is more prospering in the Expert Tasks domain
now, as the expert task domain needs expert knowledge without common sense,
which can be easier to represent and handle.
1.2.1 Characteristics of AI problems:

The problems in artificial intelligence exhibit certain traits that contribute to their
complexity:

Traits Issues

AI problems often involve incomplete or uncertain

Uncertainty information, making it difficult to determine the best
course of action.

AI problems can be highly complex, with many variables

Complexity
and interdependencies that need to be considered.

Many AI problems have multiple possible

Ambiguity interpretations, leading to ambiguity in the decision-
making process.

The relationships between variables in AI problems are

Nonlinearity often nonlinear, requiring sophisticated algorithms to
model and solve them.

The available information for AI problems is usually

Incompleteness incomplete, which can contribute to the difficulty in
finding optimal solutions.

Compiled By: P.Rajyalakshmi 10

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Noise in the data can introduce errors and make the

Noise
problem-solving process more challenging.

1.2.2 Approaches & Techniques to AI problems:

Artificial Intelligence problems span a very broad spectrum. The various approaches
and techniques provide solutions to AI problems.
 Approaches
1. Symbolic AI interprets human intelligence through symbol manipulation, rules and
logical reasoning. Basic concepts are broken down into symbols and the AI
processes them into something understandable to solve the problem.

2. Sub-symbolic approach is how an AI recognizes things or patterns without any

specific knowledge. sub symbolic AI focuses on the use of numerical representations
and machine learning algorithms to extract patterns from data. This approach, also
known as “connectionist” or “neural network”.

Compiled By: P.Rajyalakshmi 11

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

3. Statistical approach is how AI solves specific problems using mathematics tools and
statistical models such as information theory, decision theory, etc., to develop AI
algorithms. This approach has resulted in greater accuracy and reproducibility in
data mining.

 AI Technique:

AI techniques provide solutions to a variety of these problems. Artificial Intelligence

techniques refer to a “set of methods and algorithms used to develop intelligent
systems that can perform tasks requiring human-like intelligence.” These
techniques encompass the various approaches like:

o Machine Learning: This approach involves the building of algorithms to

learn patterns in data and make predictions based on it.

o Deep Learning: Deep learning uses neural networks to learn features

directly from data, making it suitable for domains with complex and
unstructured data. The main applications of deep learning can be divided into
computer vision, natural language processing (NLP).
o Natural Language Processing: Natural Language Processing involves
programming computers to process human languages to facilitate
interactions between humans and computers.
o Computer Vision: Computer Vision equips machines with the ability to
interpret visual information from the world.
o Data Mining: Data mining is the process of extracting knowledge or
insights from large amounts of data using various statistical and
computational techniques.
o Automation & Robotics: Robotic process automation is designed to carry out
high-volume, repetitive jobs while being capable of adapting to changing
conditions.

1.3 The Goals of AI:

“Artificial Intelligence (AI) is the part of computer science concerned with designing
intelligent computer systems, that is, systems that exhibit characteristics we associate
with intelligence in human behaviour – understanding language, learning, reasoning,
solving problems, and so on.” (Barr & Feigenbaum, 1981)

Compiled By: P.Rajyalakshmi 12

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Following are the two main goals of Artificial Intelligence:

 Scientific Goals: “To implement human intelligence in machines”- Creating systems

that understand, think, learn on own, and behave like humans.
 Engineering Goals (To create Expert Systems): To solve real world problems using
AI techniques such as knowledge representation, learning, rule systems, search,
and so on.

Traditionally, computer scientists and engineers have been more interested in the
engineering goal, while psychologists, philosophers and cognitive scientists have
been more interested in the scientific goal. “General intelligence is among the
field's long-term goals.”

2. The Foundation of AI:

To accomplish the traditional goals for a machine or software, Artificial Intelligence

requires the following disciplines:

1. Philosophy (the study of the fundamental nature of knowledge):

 Where does knowledge come from?
 How does knowledge lead to action?
 Aristotle (384–322 B.C.), was the first to formulate a precise set of laws
governing the rational part of the mind. He developed an informal system of
syllogisms for proper reasoning to generate valid conclusions.
E.g. all dogs are animals; all animals have four legs.
therefore, all dogs have four legs

Compiled By: P.Rajyalakshmi 13

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

2. Mathematics: Formal science requires a level of mathematical formalization in

three fundamental areas: logic, computation, and probability.
 Mathematical development really began with the work of George Boole (1815-
1864), who worked out the details of propositional or Boolean logic.
 The Italian Gerolamo Cardanao (1501-1576) first framed the idea of probability,
 Thomas Bayes (1702-1761) proposed a rule for updating probabilities to deal
with uncertain information.
 Logic and Computation: The first nontrivial algorithm is thought to be Euclid’s
algorithm for computing greatest common divisors (GCD).
3. Economics: Economics is the study of “how people make choices that lead to preferred
outcomes or utility.”
 The science of economics got its start in 1776, when Scottish philosopher Adam
Smith treats it as a science.
 Decision theory, which combines probability theory with utility theory, provides a
formal and complete framework for decisions (economic or otherwise) made under
4. Neuroscience: How does the brain work?
 Early studies (1824) relied on injured and abnormal people to understand what
parts of the brain work. More recent studies use accurate sensors to correlate
brain activity to human thought. Moore’s law states that computers will have as
many gates as humans have neurons in 2020.
5. Psychology and cognitive science: How do humans and animals think and act?
 Behaviorism movement, led by John Watson (1878-1958) discovered a lot
about rats and pigeons but had less success at understanding humans.
 Cognitive psychology views the brain as an information processing device. A
common view among psychologists is that cognitive theory should be like a
computer program. (Anderson 1980)
6. Computer engineering: How can we build an efficient computer? For artificial
intelligence to succeed, we need two things: intelligence and an artifact. The computer
has been the artifact(object) of choice.
The first operational computer was the electromechanical Heath Robinson, built
in 1940 by Alan Turing's team for a single purpose: deciphering German
messages.

Compiled By: P.Rajyalakshmi 14

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

7. Control theory and cybernetics: How can artifacts operate under their own
control?
 Machines can modify their behavior in response to the environment
(sense/action loop). Water-flow regulator, steam engine governor, thermostat,
The theory of stable feedback systems (1894)
 Weiner’s book cybernetics (1948) awoke public to the possibility of artificially
intelligent machines
8. Linguistics: How does language relate to thought?
o Modern linguistics and AI were ―born at about the same time, and grew up
together, intersecting in a hybrid field called computational linguistics or natural
language processing.

o The problem of understanding language soon turned out to be considerably

more complex than it seemed in 1957.
o Knowledge representation (the study of how to put knowledge into a form that a
computer can reason with)- tied to language and informed by research in
linguistics.
3.1 Types of Artificial Intelligence: The Strong and Weak AI

After taking a brief look at various disciplines that contribute towards AI, now let us
look at the concept of strong and weak AI which also gives foundation for developing
automated systems.

Compiled By: P.Rajyalakshmi 15

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

o Type-1(Based on Capabilities):
1. Narrow AI (Weak AI): This form of AI uses Machine learning which is
designed and trained for specific tasks or domains, such as speech
recognition, image classification, or recommendation systems. Narrow AI
excels within defined parameters but lacks general human-like intelligence. E.g.:
IBM’s Watson, Siri and Alexa are weak AI.
2. General AI (Strong AI): General AI aims to exhibit human-like
intelligence (machine intelligence) and cognitive abilities across a wide range of
tasks. This form of AI is hypothetical and remains a long-term goal of AI
research. It can find the solution to a problem and works beyond a
preprogrammed algorithm. E.g. Visual perception, speech recognition,
decision making, and translations between languages.
3. Super AI (Artificially Super AI): Super AI is AI that to go beyond in
excellence than human intelligence and ability (machine consciousness). E.g. It’s
the best at everything — maths, science, medicine, etc...,

o Type-2 (Based on functionalities):

4. Reactive machines: Reactive machines are AI systems that have no memory
and are task specific, meaning that an input always delivers the same output.
 Machine learning models tend to be reactive machines because they
take customer data, such as purchase or search history, and use it to

Compiled By: P.Rajyalakshmi 16

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

deliver recommendations to the same customers. Ex: Beat at chess by

IBM’s supercomputer Deep Blue, Netflix recommendations.
5. Limited memory machines: The next type of AI in its evolution is limited
memory. This algorithm imitates the way our brains neurons work together,
meaning that it gets smarter as it receives more data to train on.
 Deep learning algorithms improve natural language processing (NLP),
image recognition, and other types of reinforcement learning. Ex: Self-
driving cars.
6. Theory of mind: The previous two types of AI, reactive machines and limited
memory, are types that currently exist. If it is developed, theory of mind AI
could have the potential to understand the world and how other entities have
thoughts and emotions.
7. Self-awareness: The grand finale for the evolution of AI would be to design
systems that have a sense of self, a conscious understanding of their existence.
This type of AI does not exist yet.

3. History of AI:
The idea of artificial intelligence was introduced in the 1950s by Alan Mathison
Turing, a mathematician and computer scientist. Turing’s paper named “Computing
Machinery and Intelligence” included the question “Can machines think?”.

Later, in 1956, John McCarthy coined the term artificial intelligence during the
first AI conference held at Dartmouth College, New Hampshire. John McCarthy is
called “The Father of Artificial Intelligence”. Here is the history of AI during 20th
century −

Year Milestone / Innovation

Karel Čapek play named “Rossum's Universal Robots” (RUR)

1923
opens in London, first use of the word "robot" in English.

1943 Foundations for neural networks laid.

Isaac Asimov, a Columbia University alumnus, coined the

1945
term Robotics.

Compiled By: P.Rajyalakshmi 17

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Alan Turing introduced the Turing Test for evaluation of

intelligence and published Computing Machinery and
1950
Intelligence. Claude Shannon published Detailed Analysis of Chess
Playing as a search.

John McCarthy coined the term Artificial Intelligence.

1956 Demonstration of the first running AI program at Carnegie
Mellon University.

1958 John McCarthy invents LISP programming language for AI.

Danny Bobrow's dissertation at MIT showed that computers

1964 can understand natural language well enough to solve algebra
word problems correctly.

Joseph Weizenbaum at MIT built ELIZA, an interactive

1965
problem that carries on a dialogue in English.

Scientists at Stanford Research Institute Developed Shakey, a

1969 robot, equipped with locomotion, perception, and problem
solving.

The Assembly Robotics group at Edinburgh University

1973 built Freddy, the Famous Scottish Robot, capable of using vision
to locate and assemble models.

The first computer-controlled autonomous vehicle, Stanford

1979
Cart, was built.

Harold Cohen created and demonstrated the drawing

1985
program, Aaron.

Major advances in all areas of AI −

 Significant demonstrations in machine learning
 Case-based reasoning
 Multi-agent planning
1990  Scheduling
 Data mining, Web Crawler
 natural language understanding and translation
 Vision, Virtual Reality
 Games

The Deep Blue Chess Program beats the then world chess
1997
champion, Garry Kasparov.

Interactive robot pets have become commercially available. MIT

displays Kismet, a robot with a face that expresses emotions. The
2000
robot Nomad explores remote regions of Antarctica and locates
meteorites.

Compiled By: P.Rajyalakshmi 18

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

4. Sub Areas of AI:

Artificial intelligence (AI) applications are software programs that use AI
techniques to perform specific tasks.

• AI in Natural language processing (NLP): NLP allows computers to understand

and generate human language. This technology is used in a variety of applications,
such as machine translation, spam filtering, and sentiment analysis. E.g.: AltaVista's
translation of web pages. Translation of Caterpillar Truck manuals into 20 languages.
• AI in Vision Systems − These systems understand, interpret, and comprehend visual
input on the computer. For example,

 A spying Aero plane takes photographs, which are used to figure out spatial
information or map of the areas.
 Police use computer software that can recognize the face of criminal with the
stored portrait made by forensic artist.
• AI in Machine learning (ML): ML allows computers to learn from data and improve
their performance over time. This technology is used in a variety of applications, such
as predictive analytics, fraud detection, and recommendation systems.
• AI in Robotics: Robotics is the branch of AI that deals with the design, construction,
and operation of robots. Robots are used in a variety of applications, such as
manufacturing, healthcare, and space exploration.
 Self-Moving Robots: AI makes robots smart at moving around on their own.
It's like giving them a built-in GPS and a clever brain.

Compiled By: P.Rajyalakshmi 19

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

 Object Recognition and Manipulation: This is super useful, especially in

places like warehouses, where they can do things like sorting and packing
items accurately.
 AI in Expert Systems: Application-specific systems that rely on obtaining the
knowledge of human experts in an area and programming that knowledge into a
system.
 Suggestion for the spelling error while typing in the Google search box.
 MYCIN system for diagnosing bacterial infections of the blood and suggesting
treatments.
 DEC’s XCON system for custom hardware configuration. Radiotherapy treatment
planning.
NASA developed a system for classifying very faint areas in astronomical images into
either stars or galaxies.
• AI in Speech Recognition (speech-to-text recognition): It is the capacity of a
machine or program to recognize spoken words and transform them into text.
PEGASUS spoken language interface to American Airlines' EAASY SABRE
reservation system, which allows users to obtain flight information and make
reservations over the telephone.
• AI in Mathematical Theorem Proving Use inference methods to prove new
theorems.
• AI in Scheduling and Planning: Dynamic analysis and replanning tool (DART)used
in Desert Storm and Desert Shield operations performs automated planning and
scheduling for transportation. European space agency planning and scheduling of
spacecraft assembly, integration and verification.
5. Applications of AI
Applications of AI in a wide variety of industries including healthcare, finance,
manufacturing etc. are as follows:

AI in Astronomy: Machines can assist with data processing in the astronomical sector,
such as capturing new stars, extraterrestrial planets, and even dark matter.

AI in Gaming:AI plays crucial role in strategic games such as chess (Deep Blue Chess
program beat world champion Gary Kasparov), poker, tic-tac-toe, etc., where machine can
think of large number of possible positions based on heuristic knowledge.

Compiled By: P.Rajyalakshmi 20

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

AI in Healthcare: AI is used for medical diagnosis, drug discovery, and predictive

analysis of diseases.

 AI in Finance: AI helps in credit scoring, fraud detection, and financial forecasting.

 AI in Retail: AI is used for product recommendations, price optimization, and
supply chain management.
 AI in Manufacturing: AI helps in quality control, predictive maintenance, and
production optimization.
 AI in Transportation: AI is used for autonomous vehicles, traffic prediction, and
route optimization.
 AI in Automotive Industry: AI is being used to develop self-driving cars. Advanced
Driver Assistance Systems (ADAS) autonomously adjust your vehicle's speed while on
the highway, swiftly engage the brakes when detecting potential hazards.
 AI in Social Media: Virtual Assistants and AI Chatbots on social media are used for
customer support, answering frequently asked questions, and handling simple
requests. AI-powered Sentiment Analysis tools can figure out how people feel (happy,
sad, or neutral) on social media based on their comments or posts in message.
 AI in Entertainment: Recommendation of Content: AI looks at what customers have
liked before, such as movies or music, and suggests new things that they might enjoy.
 AI in Agriculture: (Crop Observation and Control) AI, with the help of various sensors,
acts as a guardian for crops on the farm. AI controls several machines like tractors
and drones. These machines can plant seeds, remove weeds, and spray stuff on crops
all by themselves.
 AI in E-commerce: Personalized Product recommendations, dynamic pricing, virtual
shopping assistants etc.,

Compiled By: P.Rajyalakshmi 21

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Review Questions:
Part A – Two Marks
1. Distinguish Intelligence and Artificial Intelligence. L1, CO1
2. What is A.I. technique? Also tell the areas where it can be used. L2, CO1
3. What are the goals of AI? L1,CO1 Refer 1.2
4. List the advantages and disadvantages of AI?L1,CO1 Refer 10
5. What is meant by Turing test? Refer 1.2
Part B -Ten Marks
1. Define AI? Explain the organization of AI. L2, CO1 Refer 1,1.1
2. Describe the Turing test for machine intelligence. Highlights its important features.
L2, CO1 Refer 1.1
3. Explain about the foundations of AI. L2, CO1 Refer 2
4. Classify AI problems and Discuss the characteristics of AI Problems.L2, CO1 Refer
1.2,1.2.2
5. Write a short note on history of AI along with milestones. L2, CO1 Refer 3

Compiled By: P.Rajyalakshmi 22

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Chapter 2-Intelligent Agents

1. Agents and Environment:

An AI system is composed of an agent and its environment. The agents act in their
environment. An environment in artificial intelligence is the surrounding of the
agent.

So, we define-

Agent as “an independent program or an autonomous entity that is designed to perceive its
environment through sensors and acts upon that environment through actuators”.

Agents run through a cycle of perception, thought, and action.

Figure: Agents interact with environments through sensors and actuators.

Compiled By: P.Rajyalakshmi 23

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Examples of an agents include:

 A human agent has sensory organs such as eyes, ears, nose, tongue and skin parallel
to the sensors, and other organs such as hands, legs, mouth, for actuators and physical
world as environment.
 A robotic agent can have cameras, infrared range finder, NLP for sensors and various
motors for actuators, physical world as environment.
 A software agent can have keystrokes, file contents as sensory input and act on those
inputs and display output on the screen. Operating system as environment.

1.1 Agent Terminology:

o Performance Measure of Agent − It is the criteria which determines how
successful an agent is.
o Behavior of Agent − It is the action that agent performs after any given
sequence of percepts.
o Percept − It is agent’s perceptual inputs at a given instance.
o Percept Sequence − It is the history of all that an agent has perceived till
date.
o Agent Function − It is a map from the precept sequence to an action.
f: P* → A
o Agent Program: A software/algorithm where agent function is typically
implemented.
o Intelligent Agent: An intelligent agent is an autonomous entity which act
upon an environment using sensors and actuators for achieving goals.

2. The Concept of Rationality:

In AI, rational agents are closely related to intelligent agents or autonomous

programs that mimic human intelligence. It operates under the premise of rationality.

Compiled By: P.Rajyalakshmi 24

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Rationality can be judged based on following points:

 The existence of a performance measure that defines the success criterion.

 The agent having prior knowledge about its environment.

 The best possible actions for an agent to perform. (actuator dependency)
 The agent’s percept sequence

2.1 Rational Agent

Definition- “A rational agent is an agent which has clear preference, models uncertainty, and acts
in a way to maximize its performance measure with all possible actions.”

A rational agent is said to perform the right things. A rational agent is anything
that

• Perceive its environment through sensors

• Think and make decisions based on logical reasoning and
• Act through actuators

AI is about creating rational agents to use for game theory and decision theory for
various real-world scenarios.

Compiled By: P.Rajyalakshmi 25

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

2.2 P.E.A.S Representation:

A task environment is a “problem” to which rational agent is a solution. In
designing a rational agent, we need to specify the task environment by PAGE
(Percept, Action, Goal, Environment) or P.E.A.S Representation.

 Performance Measure It defines the success of an agent. It evaluates the criteria

that determines whether the system performs well.

 Environment refers to the physical or virtual surroundings and conditions in

which the AI system operates.

 Actuators These are the mechanisms through which an AI agent interacts with its
environment, such as motors, wheels, or speakers.

 Sensors These devices enable an AI agent to perceive its surroundings, gathering

necessary data for decision-making, like cameras, microphones, or thermometer.

3. The Nature of Environments:

An environment in artificial intelligence is the surrounding of the agent. The

agent takes input from the environment through sensors and delivers the output to
the environment through actuators. The environment is mostly said to be non-
feministic.

Compiled By: P.Rajyalakshmi 26

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

3.1 Properties of Environment:

The environment in Artificial Intelligence refers to the outside circumstances

in which an agent functions to accomplish a particular task. There are several
types of environments:

1. Fully observable (vs. partially observable): A fully observable environment

is one in which the agent has complete information about the current state
of the environment. The agent has direct access to all environmental features
that are necessary for making decisions.
Example: board games like chess or checkers- the board is fully observable,
and so are the opponent’s moves.

 A partially observable environment is one in which the agent does not have
complete information about the current state of the environment. The agent
can only observe a subset of the environment, and some aspects of the
environment may be hidden or uncertain.
 Example: driving a car in traffic. The environment is partially observable
because what’s around the corner is not known.

2. Deterministic (vs. stochastic): We say the environment is deterministic if

the next state of the environment is completely determined by the
current state and the action executed by the agent.
Example: simple mathematical equations, where the outcome of each
operation is precisely defined.

Compiled By: P.Rajyalakshmi 27

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

 A stochastic environment is one in which the outcome of an action is

uncertain and involves probability.
 Examples: games of chance like poker or roulette, where the outcome of
each action is influenced by random factors like the shuffle of cards or the
spin of a wheel.
3. Episodic (vs. sequential): In an Episodic task environment, each of the
agent’s actions is divided into atomic incidents or episodes. There is no
dependency between current and previous incidents. In each incident, an
agent receives input from the environment and then performs the
corresponding action.
Example: Consider an example of Pick and Place robot, which is used to
detect defective parts from the conveyor belts. Here, every time robot(agent)
will make the decision on the current part i.e. there is no dependency between
current and previous decisions.

 In a Sequential environment, the current or previous decisions can

affect all future decisions. The next action of the agent depends on what
action he has taken previously and what action he is supposed to take in the

Compiled By: P.Rajyalakshmi 28

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

future. Example: Checkers- Where the previous move can affect all the
following moves.
4. Static (vs. dynamic): is one in which the environment does not change over
time. The state of the environment remains constant, and the agent's actions
do not affect the environment. Ex: Cross word puzzles

A dynamic environment is one in which the environment changes over time

and the agent's actions can affect the future state of the environment.
Examples include video games or robotics applications.

5. Discrete (vs. continuous): If an environment consists of a finite number of

actions that can be deliberated in the environment to obtain the output, it is
said to be a discrete environment. Example: The game of chess is discrete
as it has only a finite number of moves though the number of moves might
vary.
In contrast, a continuous environment is one in which the state and action
spaces are continuous and infinite. Ex: robotics or control systems.
6. Single agent (vs. multiagent): An environment consisting of only one
agent is said to be a single-agent environment. Ex: An agent solving a

crossword puzzle.
An environment involving more than one agent is a multi-agent
environment. Examples: multiplayer games, traffic simulations.

4. Structure of an AI Agent:
The task of AI is to design an agent program which is an implementation of the
agent function. Structure of an Intelligent Agent is a combination of agent,
architecture and agent program and can be viewed as:

Agent = Architecture + Agent Program

Architecture is a device with sensors and actuators that the agent executes on. for
example, a robotic car, a camera, and a PC.
Agent function is a map from the percept sequence to an action which takes the
entire percept history.

Compiled By: P.Rajyalakshmi 29

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Agent program is an implementation of the agent function which executes on a

computing device called architecture and takes the current percept as input from
sensors and return action to the actuators.
A) Schematic of Agent’s Structure B) Role of agent architecture in agent program

4.1 Types of Intelligent Agents in AI Environment:

Here are the five main types of AI agents:

1. Simple reflex agents are programmed to only succeed in the fully observable
environment and take decisions based on the current precepts and ignore the
rest of the percept history.
 The Simple reflex agent works on Condition-action rule, which means it maps
the current state to action.
 Such as a Room Cleaner agent, it works only if there is dirt in the room.
which means it maps the current state to action. Such as a Room Cleaner agent,
it works only if there is dirt in the room.

Compiled By: P.Rajyalakshmi 30

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Problems for the simple reflex agent design approach:

 They have very limited intelligence

 They do not have knowledge of non-perceptual parts of the current state
 Mostly too big to generate and to store.
 Not adaptive to changes in the environment.

2. Model-based reflex agents can work in a partially observable environment

and track the situation. A model-based agent has two important factors:
Model: It is knowledge about "how things happen in the world," so it is called a
Model-based agent.
Internal State: It is a representation of the current state based on percept
history. Updating the agent state requires information about:
 How the world evolves independently from the agent?
 How do the agent’s actions affect the world?

Compiled By: P.Rajyalakshmi 31

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

3. Goal-based agents execute a program to achieve specific goals and take

actions based on evaluating the current state of the environment. These agents
may have to consider a long sequence of possible actions before deciding
whether the goal is achieved or not. They usually require search and planning.
The goal-based agent’s behavior can easily be changed.

4. Utility-based agents consider the potential outcomes of their actions and

choose the best way to achieve the goal. The utility function maps each state
to a real number to check how efficiently each action achieves the goals. They
choose actions based on a preference (utility) for each state.

Compiled By: P.Rajyalakshmi 32

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

5. Learning agents execute machine learning techniques to improve their

decision-making over time.

 A learning agent in AI is the type of agent that can learn from its past
experiences, or it has learning capabilities.

 It starts to act with basic knowledge and then can act and adapt automatically
through learning.

A learning agent has mainly four conceptual components, which are:

a. Learning element: It is responsible for making improvements by

learning from the environment.
b. Critic: The learning element takes feedback from critics which
describes how well the agent is doing with respect to a fixed
performance standard.
c. Performance element: It is responsible for selecting external action.
d. Problem Generator: This component is responsible for suggesting
actions that will lead to new and informative experiences.

4.2 Problem Solving Agents in AI

A problem-solving agent is a goal-based agent and focuses on achieving goals. It

decides what to do by finding sequence of actions that lead to desirable states.

Compiled By: P.Rajyalakshmi 33

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

1) Goal formulation define the problems that the agents need to solve.
2) Problem formulation is the process of deciding what actions and states to
consider, given a goal.

The five components involved in problem formulation:

i. Initial State: This state requires an initial state for the problem which
starts the AI agent towards a specified goal. In this state new methods also
initialize problem domain solving by a specific class.
ii. Action: This stage of problem formulation works with function with a
specific class taken from the initial state and all possible actions done in this
stage.
iii. Transition: This stage of problem formulation integrates the actual action
done by the previous action stage and collects the final stage to forward it to
their next stage.
iv. Goal test: This stage determines that the specified goal achieved by the
integrated transition model or not, whenever the goal achieves stop the
action and forward into the next stage to determines the cost to achieve the
goal.
v. Path costing: This component of problem-solving numerical assigned what
will be the cost to achieve the goal. It requires all hardware, software and
human working cost.
3) Search: The process of looking for a sequence of actions that reaches the goal is
called search. A search algorithm takes a problem as input and returns a solution
in the form of an action sequence.

Compiled By: P.Rajyalakshmi 34

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

4) Solution: It finds the best algorithm out of various algorithms, which may be
proven as the best optimal solution. Finding a solution of a problem involves
following phases:
a. Problem definition: Detailed specification of inputs and acceptable
system solutions.
b. Problem analysis: Analyse the problem thoroughly.
c. Knowledge Representation: collect detailed information about the
problem and define all possible techniques.
d. Problem-solving: Selection of best techniques.
5) Execution phase: Once a solution is found, the carrying actions it recommends is
called the execution phase.

A. Toy Problems
1. 8-Puzzle Problem

States: A state of the 8-puzzle is a 3x3 grid, with eight numbered tiles and a blank
space. The objective is to reach the goal state as shown in Figure. We can slide four
adjacent tiles (left, right, above, and below) into the empty space.

Compiled By: P.Rajyalakshmi 35

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

The problem formulation is as follows:

• Initial State: We can start from any state as the initial state.
• Actions: Here, actions or the movement of the blank space is defined, i.e., either
left, right, up or down. The empty space cannot move diagonally and can take
only one step at a time.
• Transition Model: Given the state and action it returns the resulting state.
• Goal test: It identifies whether we have reached the correct goal-state.
• Path cost: The path cost is the number of steps in the path where the cost of each
step is 1.

Note: The 8-puzzle belongs to the family of sliding-block puzzles.

State Space for 8-Puzzle problem

2. Vacuum World
States: The state is determined by both the agent location and the dirt locations. The
agent is in one of two locations, each of which might or might not contain dirt. Thus,
there are 2 × 2^2 = 8 possible world states.

The problem formulation is as follows:

• Initial state: Any state can be designated as the initial state.

• Actions: Each state has just three actions: Left, Right, and Suck.
• Transition model: The actions have their expected effects, except that moving
Left in the leftmost square, moving Right in the rightmost square, and Sucking in
a clean square have no effect. The transition model defines a state space.

• Goal test: This checks whether all the squares are clean.
• Path cost: Each step costs 1, so the path cost is the number of steps in the path.
State space for the vacuum world
 Links denote actions: L = Left, R =Right, S = Suck.

Compiled By: P.Rajyalakshmi 36

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

3. 8- Queens problem
The goal of the 8-queens problem is to place 8 queens on the chessboard such that no
queen attacks any other. (A queen attacks any piece in the same row, column or
diagonal).

For this problem, there are two main kinds of formulation:

1. Incremental formulation: It starts from an empty state where the operator
augments a queen at each step.
Following steps are involved in this formulation:

States: Arrangement of any 0 to 8 queens on the chessboard.

Initial State: An empty chessboard
Actions: Add a queen to any empty box.
Transition model: Returns the chessboard with the queen added in a box.
Goal test: Checks whether 8-queens are placed on the chessboard without any attack.
Path cost: There is no need for path cost because only final states are counted.
In this formulation, there is approximately 1.8 x 1014 possible sequence to
investigate.

Compiled By: P.Rajyalakshmi 37

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

2. Complete-state formulation: It starts with all the 8-queens on the chessboard and
moves them around, saving from the attacks.
Following steps are involved in this formulation
States: Arrangement of all the 8 queens one per column with no queen attacking the
other queen.
Actions: Move the queen at the location where it is safe from the attacks.
This formulation is better than the incremental formulation as it reduces the state
space from 1.8 x 1014 to 2057, and it is easy to find the solutions.
B. Real-World Problems
1. Route-finding problems are defined in terms of specified locations and
transitions along links between them.
 E.g. Traveling salesperson problem (TSP): It is a touring problem where
the salesman can visit each city only once. The objective is to find the
shortest tour and sell-out the stuff in each city.
 Route-finding algorithms are used in a variety of applications such as Web
sites and in-car systems that provide driving directions.
2. VLSI layout problem requires positioning millions of components and
connections on a chip to minimize area, minimize circuit delays, minimize stray
capacitances, and maximize manufacturing yield.

3. Air Line Travel Problem

The airline travel problem is specifying as follows:
1. States: Each is represented by a location (e.g., an airport) and the current time.
2. Initial state: This is specified by the problem.
3. Transition state: This returns to the states resulting from taking any
scheduled flight (further specified by seat class and location), leaving later than the
current time plus the within-airport transit time, from the current airport to
another.
4. Goal Test: Are we at the destination by some prespecified time?
5. Path cost: This depends upon the monetary cost, waiting time, flight time,
customs and immigration procedures, seat quality, time of Date, type of airplane,
frequent-flyer mileage awards, and so on.

Compiled By: P.Rajyalakshmi 38

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Review Questions:

Unit-I

Part A - Two Marks

1. Define Intelligent Agent.L1, CO1

2. Distinguish an agent function and an agent program. L2, CO1
3. State the concept of Rationality? L1, CO1 Refer 2.
4. Define Rational agent? L1, CO1 Refer 2.1
5. What is a task environment? How is it specified? L1, CO1 Refer 3.1
6. Give the structure of an agent in an environment? L1, CO1 Refer 4
7. What are the four different kinds of agent programs? L2, CO1 Refer 4.1
Part B – Ten Marks
1. Elaborate in detail about reflex and goal-based agents. Illustrate the use of learning agents.
L2, CO1 Refer 4.1
2. List the properties of task environment? Explain in detail about each property with
examples. L2, CO1 Refer 4
3. What is a simple problem-solving agent? Explain it briefly. L2, CO1 Refer 4.2
4. What is problem formulation? Formulate 8 puzzle problem as AI problem. L3, CO1
Refer 4.2,4.2(A)
5. Explain the applications of AI? L2, CO1 Refer 5

Compiled By: P.Rajyalakshmi 39

S.V.P.C.E.T
 Department of CSE(AI&ML) II Year B. Tech-I Semester

UNIT - II

Searching- Searching for solutions, uniformed search strategies – Breadth first search, depth
first Search. Search with partial information (Heuristic search) Hill climbing, A*, AO*
Algorithms, Problem reduction, Game Playing-Adversial search, Games, mini-max algorithm,
optimal decisions in multiplayer games, Problem in Game playing, Alpha-Beta pruning,

1. Searching:
Searching is a step-by-step procedure to solve a search-problem in search space.

a. Searching for solutions:

“Solution to a search problem is a sequence of actions, that transforms the start state to the goal state”.
This plan is achieved through search strategies or algorithms.

b. Search Algorithm:

The search algorithm takes “a problem as input and returns a solution in the form of action
sequence”. Each of these algorithms will have:

A problem graph, containing the start node S and the goal node G.

A strategy, describing the way the graph will be traversed to get to G.

A frontier(fringe) is a data structure Queue (FIFO), or Stack (LIFO) used to store all the
possible states (nodes) that you can go from the current states.

c. Search Tree:

A tree representation of search problem is called Search tree. The root of the search tree is
the root node which corresponds to the initial state.

For each Node of the tree, we have a structure that contains four components:

• State: The state in the state space to which the node corresponds.
• Parent node: The node in the search tree that generated this node.
• Action: The action that was applied to the parent to generate the node.
• Path-cost: The cost, traditionally denoted by g(n), of the path from the initial state
to the node, as indicated by the parent pointers.
• Depth: The number of steps along the path from initial state.

Compiled By: P.Rajyalakshmi 40

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

d. Properties of Search Algorithms:

The Criteria to measure the performance of different search strategies:

Completeness: A search algorithm is said to be complete if it guarantees to return a solution if

at least any solution exists for any random input.

Optimality: If a solution found for an algorithm is guaranteed to be the best solution (lowest
path cost) among all other solutions, then such a solution for is said to be an optimal solution.

Time Complexity: Time complexity is a measure of time for an algorithm to complete its task.

Space Complexity: It is the maximum storage space required at any point during the search, as
the complexity of the problem.

2. Types of search algorithms

Based on the search problems we can classify the search algorithms into uninformed (Blind
search) search and informed search (Heuristic search) algorithms.

2.1 Uninformed (Blind) Search Strategies

A way in which search tree is searched without any information about the search space like initial
state operators and test for the goal, so it is also called blind search.
It can be divided into six main types:

Compiled By: P.Rajyalakshmi 41

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

1. Breadth-first Search
2. Depth-first Search
3. Depth-limited Search
4. Iterative deepening depth-first search
5. Uniform cost search
6. Bidirectional Search
a. Breadth-First Search Algorithm
 BFS algorithm starts searching from the root node of the tree and expands all
successor node at the current level before moving to nodes of next level.
 It always expands the Shallow nodes first.
 BFS is implemented by calling TREE-SEARCH (problem, FIFO-QUEUE ())
 Queue data structure puts all newly generated successors at the end of the queue.

Path: A-B-C-D-E-F-G-H-I-J-K

Advantages:

 BFS guarantees the shortest path to the goal in terms of the number of actions.
 It's also complete, meaning it will find a solution if one exists.

Limitations:

 It can be memory-intensive, especially in large state spaces.

 It may not be efficient for searching deep or infinite spaces.

Properties:

Compiled By: P.Rajyalakshmi 42

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

b. Depth First Search:

• Depth-first search is a recursive algorithm for traversing a tree or graph data

structure.
• It always expands to the deepest node in the frontier of the search tree.
• DFS is implemented in TREE-SEARCH using a last-in first-out (LIFO) queue data
structure.
• Generated states are added at the front of the Queue.
• Root node--->Left node ----> right node

Compiled By: P.Rajyalakshmi 43

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Advantages:

It is memory-efficient and well-suited for deep or infinite state spaces.

It can find a solution quickly if it's not too far from the initial state.

Limitations: It does not guarantee the shortest path, and it may get stuck in infinite loops
or deep branches

Properties:

2.2 Search with partial information:

If environment is not fully observable and knowledge of the states and actions is incomplete,
then agent has different types of task environment. Such an environment leads to three
distinct types of problems -

• Sensor less or conformant problem - Agent may have no sensors i.e. no idea where it
is; solution is a sequence.

• Contingency problem – Percepts provide new information about current state;

solution is a tree or policy; often interleave search and execution. A contingency

Compiled By: P.Rajyalakshmi 44

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

problem is called as adversarial problem if the uncertainty is caused by the actions of

another agent.

• Exploration problem – When states and actions of the environment are unknown.

2.3 Heuristic search:

In an informed search, problem information is available which can guide the search. Informed
search is also called a Heuristic search.

Informed search strategies can find a solution more efficiently than an uninformed search
strategy.

They are:

1. Hill Climbing
2. A* Search
3. AO* search

An example of informed search algorithms is a traveling salesman problem.

Heuristic function: It is represented by h(n), and it calculates the cost of an optimal path
from the current state to goal state. The value of the heuristic function is always positive.

2.3.1. A* Search Algorithm

The A* (A-star) algorithm is a popular pathfinding and graph traversal algorithm used to
find the shortest path between two nodes in a graph. It employs a “heuristic estimate” as
well as the cost to reach the node. Hence, we can combine both costs as follows, and this
sum is called a fitness number f(n).

Algorithm of A* search:

Step1: Place the starting node in the OPEN list.

Step 2: Check if the OPEN list is empty or not, if the list is empty then return failure and
stops.

Step 3: Select the node from the OPEN list which has the smallest value of evaluation
function (g+h), if node n is goal node, then return success and stop, otherwise

Compiled By: P.Rajyalakshmi 45

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Step 4: Expand node n and generate all its successors and put n into the closed list. For
each successor n', check whether n' is already in the OPEN or CLOSED list, if not then
compute evaluation function for n' and place into Open list.

Step 5: Else if node n' is already in OPEN and CLOSED, then it should be attached to the
back pointer which reflects the lowest g(n') value.

Step 6: Return to Step 2.

Advantages: A* guarantees an optimal solution (shortest path) if an admissible heuristic is

used. It's also efficient in terms of the number of nodes expanded.

Limitations: The quality of the heuristic matters. If the heuristic is not admissible, the
optimality guarantee is lost.

Applications: Tower Défense is a type of strategy video game where the goal is to defend a
player’s territories or possessions by obstructing enemy attackers, usually achieved by
placing defensive structures on or along their path of attack.

Example:
In this example, we will traverse the given graph using the A* algorithm. The heuristic
value h (n) of all states is given in the below table. Here we will use OPEN and CLOSED
list.

Initialization: {(S, 5)}

Iteration1: {(S--> A, 4), (S-->G, 10)}

Compiled By: P.Rajyalakshmi 46

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Iteration2: {(S--> A-->C, 4), (S--> A-->B, 7), (S-->G, 10)}

Iteration3: {(S--> A-->C--->G, 6), (S--> A-->C--->D, 11), (S--> A-->B, 7), (S-->G, 10)}

Iteration 4 will give the result, as S--->A--->C--->G

it provides the optimal path cost=6

2.3.2 Hill-Climbing:

The Hill-Climbing search algorithm is simply a loop that continually moves in the direction
of increasing value – that is, uphill.

It terminates when it reaches a “peak” where no neighbor has a higher value.

The idea behind hill climbing is as follows:

1. Pick a random point in the search space.

2. Consider all the neighbors of the current state.

3. Choose the neighbor with the best quality and move to that state.

4. Repeat 2 thru 4 until all the neighbouring states are of lower quality.

5. Return the current state as the solution state.

Algorithm:
Function HILL-CLIMBING(Problem) returns a solution state Inputs: Problem, problem
Local variables: Current, a node
Next, a node
Current = MAKE-NODE(INITIAL-STATE[Problem])
Loop do
Next = a highest-valued successor of Current
If VALUE[Next] < VALUE[Current] then return Current
Current = Next
End

Problems with Hill Climbing:

• The main problem with hill climbing (which is also sometimes called gradient
descent) is that we are not guaranteed to find the best solution.
• Hill Climbing can get stuck in local optima, which are suboptimal solutions.
• The choice of the initial solution can significantly impact the algorithm's
performance.

Compiled By: P.Rajyalakshmi 47

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Applications: Hill Climbing finds applications in network routing, hyperparameter tuning,

game strategy optimization, chip design, and more.

2.4 Problem Reduction

• Problem reduction search is a basic problem-solving technique of AI.

• A problem can be divided into a set of sub problems, where each sub problem can be
solved separately and a combination of these will be a solution.

• The AND-OR (AO*) graphs are used for representing the solution.

2.4.1 AO* algorithm

The AO* method divides any given difficult problem into a smaller group of problems
that are then resolved using the AND-OR graph concept.

The AND side of the graph represents a set of tasks that must be completed to achieve the
main goal, while the OR side of the graph represents different methods for accomplishing the
same main goal.

The evaluation function in AO*: f(n) = Actual cost + Estimated cost

f(n) = g(n) + h(n)

Compiled By: P.Rajyalakshmi 48

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

2.4.1 Game Playing -Adversarial search

Searches in which two or more players with conflicting goals are trying to explore the same
search space for the solution, are called adversarial searches, often known as Games.

Games are modelled as a Search problem and heuristic evaluation function, and these are the
two main factors which help to model and solve games in AI.

Formalization of the problem:

A game can be formally defined as a search problem with the following components:

• The initial state, which includes the board position and identifies the player to move.

• A successor function, which returns a list of (move, state) pairs.

• A terminal state-States where the game has ended.

• A utility function (also called an objective function or payoff function), which give a
numeric value for the terminal states.

2- Person Games

Players: We call them Max and Min.

Initial State: Includes board position and whose turn it is.

Operators: These correspond to legal moves.

Compiled By: P.Rajyalakshmi 49

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

Terminal Test: A test applied to a board position which determines whether the game is
over. In chess, for example, this would be a checkmate or stalemate situation.

Utility Function: A function which assigns a numeric value to a terminal state. For
example, in chess the outcome is win (+1), lose (-1) or draw (0).

Note that by convention, we always measure utility relative to Max.

2.4.2 Mini-max Algorithm

1. Generate the whole game tree.

2. Apply the utility function to leaf nodes to get their values.

3. Use the utility of nodes at level n to derive the utility of nodes at level n-1.

4. Continue backing up values towards the root (one layer at a time).

5. Eventually the backed-up values reach the top of the tree, at which point Max
chooses the move that yields the highest value. This is called the minimax decision
because it maximizes the utility for Max on the assumption that Min will play
perfectly to minimize it.

Compiled By: P.Rajyalakshmi 50

S.V.P.C.E.T
Department of CSE(AI&ML) II Year B. Tech-I Semester

2.4.3 Alpha-beta pruning

Alpha-beta pruning is an optimization technique for the minimax algorithm. By performing
pruning, we can eliminate large part of the tree from consideration.

α: the value of the best (i.e., highest value) choice we have found so far at any choice point
along the path of MAX.

β: the value of best (i.e., lowest value) choice we have found so far at any choice point along
the path of MIN.

The main condition which required for alpha-beta pruning is: α>=β

Key points about alpha-beta pruning:

• The Max player will only update the value of alpha.

• The Min player will only update the value of beta.

• While backtracking the tree, the node values will be passed to upper nodes instead of
values of alpha and beta.

• We will only pass the alpha, beta values to the child nodes.

Compiled By: P.Rajyalakshmi 51

S.V.P.C.E.T
 Department of CSE(AI&ML) II Year B. Tech-I Semester

Review Questions

Unit-II

Part-A

Distinguish uninformed and informed search strategies?

Define heuristic functions
List the criteria to measure the performance of search strategies.
What are the advantages of Breadth-First Search.
Define Search tree and write its properties.
What is AO* Search?
What is game tree?
Define Evaluation function.
Part-B

Describe searching for solutions. Give the structure of search tree.

Elucidate search algorithms with uninformed search strategies.

Define Heuristic search? Explain any two Heuristic Search Techniques

Write in detail the Breadth-First Search algorithm. Illustrate with example

Compiled By: P.Rajyalakshmi 52

S.V.P.C.E.T
 Department of CSE(AI&ML) II Year B. Tech-I Semester

Explain about A* search in detail.

Illustrate the heuristic Hill climbing algorithm with an example.

What is problem reduction. Illustrate AO* search with example.

Describe Adversial Search in game-playing with an example.

Explain Minmax search in gaming.

Elaborate in detail about alpha-beta pruning with example.

Compiled By: P.Rajyalakshmi 53

S.V.P.C.E.T
 Department of CSE (AI & ML) II Year B.Tech-I Semester

LECTURE PLAN
UNIT III

S.No. Topic No. of Proposed Actual Pertaining Taxonomy Mode of

Periods lecture lecture CO(s) Level Delivery

Period Period
1 Knowledge
Representation 1 CO3 L2 MD1,
issues MD5
2 Predicate logic MD1,
1 CO3 L1 MD5
3 Logic 1 MD1,
Programming CO3 L2 MD5

4 Semantic Nets
1 CO3 L1 MD1,
MD5
5 Frames and 1 CO3 L1 MD1,
Inheritance MD5

6 Constraint 1 CO3 L2 MD1,

Propagation MD5
7 Representing
knowledge 1 CO3 L2 MD1,
using Rules MD5
8 Rules-based 1
deduction CO3 L2 MD1,
systems MD5

9 Reasoning 1
under CO3 L3 MD1,
uncertainty MD5
10 Review of 1 CO3 L1 MD1,
probability MD5
11 Bayes’ 1 CO3 L1 MD1,
probabilistic MD5
interferences
12 Dempster 1 CO3 L1 MD1,
Shafer theory MD5

1
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

UNIT - III
Representation of Knowledge: Knowledge representation issues, predicate logic- logic
programming, semantic nets- frames and inheritance, constraint propagation, representing
knowledge using rules, rules-based deduction propagation, representing knowledge using
rules, rules-based deduction systems. Reasoning under uncertainty, review of probability,
Bayes’ probabilistic interferences and dempster Shafer theory.

1. Introduction

Knowledge Representation in AI describes the representation of knowledge in a structured

form. The fundamental goal of knowledge representation is to make inferences and draw
conclusions. To achieve this, various knowledge representation techniques can be used, such
as logical representation, semantic network representation, frame representation, and production
rules.

As we can see in above diagram, there is one decision maker which act by sensing the
environment and using knowledge. But if the knowledge part will not present then, it
cannot display intelligent behavior.

Any knowledge representation system should possess the properties such as learning,
efficiency in acquisition, representational adequacy and inferential adequacy.

Knowledge representation in AI is not just about storing data in a database, it allows a

machine to learn from that knowledge and behave intelligently like a human being.

For example, a knowledge representation system might be used to build a chatbot that
can answer questions about a particular topic or a recommendation engine that can suggest
products based on a user's preferences.

2
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

1.1 Cycle of Knowledge Representation in AI

Artificial Intelligent Systems usually consist of various components to display their

intelligent behavior. Some of these components include:

 Perception

 Learning

 Knowledge Representation & Reasoning

 Planning

 Execution

1.2 What to Represent:

The knowledge that needs to be represented in AI are objects, events, performance, facts,
meta-knowledge, and Knowledgebase.

1. Objects: Objects refer to things in the world that have physical properties and can
be observed, touched, or manipulated. Object-oriented programming is an example of a
technique that uses objects to represent knowledge in AI
2. Events: Events refer to actions or occurrences that take place in the world. Event-
based systems use events to represent knowledge in AI. Examples of events include
driving a car, cooking food.

3
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

3. Performance: Performance refers to the behavior of agents or systems that perform

a task. Performance-based systems use performance to represent knowledge in AI.
4. Facts refer to propositions that are either true or false. They are statements that
can be verified using evidence or logical deduction. Example: "the sky is blue,"
5. Meta-Knowledge refers to knowledge about knowledge. It helps machines reason
about the quality and validity of the knowledge they are using.
6. Knowledge Base: A knowledge base is a collection of facts, rules, procedures, and
another knowledge relevant to a particular domain.
1.3 Properties of Knowledge Representation
A good knowledge representation system must possess the following properties:
* Representational Accuracy: KR system should have the ability to represent all
kind of required knowledge.
* Inferential Adequacy: KR system should have ability to manipulate the
representational structures to produce new knowledge corresponding to existing
structure.
* Inferential Efficiency: The ability to direct the inferential knowledge mechanism
into the most productive directions by storing appropriate guides.
* Acquisitional efficiency: The ability to acquire the new knowledge easily using
automatic methods.
1.4 Knowledge Representation Issues
1. Incompleteness and Uncertainty: Incompleteness and uncertainty can lead to
incorrect diagnoses or treatment recommendations.
2. Scalability Problems: Scalability issues can result in slow response times and
hamper the performance of AI-driven recommendation systems.
3. Representation Bias: Representation bias occurs when knowledge representations
favour certain perspectives or groups, leading to unfair or inaccurate decision-
making.

4
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

4. Handling Context and Context-Dependent Knowledge: AI systems must be able to

handle context-dependent knowledge to provide accurate guidance.
5. The Symbol-Grounding Problem: The symbol-grounding problem is concerned
with connecting abstract symbols in AI systems to real-world entities.
6. Ontology Development and Maintenance Challenges: Ontologies are fundamental
for knowledge representation, but developing and maintaining them can be complex.

1.5 Knowledge Representation Techniques

The Knowledge Representation models/mechanisms are often based on:

1) Logical Representation: This method employs formal logic to represent information

systematically.
2) Semantic Networks: These graphical representations connect related concepts,
making it easier to visualize relationships.
3) Production Rules: Utilized to establish a set of rules for decision-making and
problem-solving.
4) Frames Representation: This technique structures information into frames or
templates, enhancing data organization.
1.5.1 Logical Representation
Logical representation means drawing a conclusion based on various conditions. Each
sentence can be translated into logics using syntax and semantics.
* Syntax: Syntaxes are the rules which decide how we can construct legal
sentences in the logic. It determines which symbol we can use in knowledge
representation and how to write those symbols.
* Semantics: Semantics are the rules by which we can interpret the sentence in the
logic. Semantic also involves assigning a meaning to each sentence.
 Advantages:

5
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

* Logical representation helps to perform logical reasoning.

* This representation is the basis for the programming languages.
 Disadvantages:
* Logical representations have some restrictions and are challenging to work
with.
* This technique may not be very natural, and inference may not be very
efficient.
1.5.2 Approaches of Logical Representation
Logical representation can be categorised into mainly two logics:
1. Predicate Logic
2. Propositional Logic

A. Predicate Logic in AI
Predicate logic in artificial intelligence, also known as first-order logic (FOL), is a
formal system used in logic and mathematics to represent and reason about complex
expressions in easier forms using predicates, variables, and quantifiers. It can represent
negation, conjunction, disjunction, and many more types of statements by using
symbols.

Figure: Logic symbols used in Predicate Logic

 Basic components of predicate logic are:
* Predicates are statements or propositions that can be either true or false depending on
the values of their arguments.
* Variables are symbols that can take on different values. In predicate logic, variables
are used to represent objects or entities in the domain of discourse.
* Constants are specific values that do not change. For instance, in a knowledge base
about people, "Alice" and "Bob" might be constants representing specific
individuals.

6
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

* Quantifiers are used to specify the scope of variables in logical expressions. There are
two types of quantifiers in predicate logic –
a) Universal Quantifier (∀): it indicates that the statement within the quantifier
is true for all objects in the domain
b) Existential Quantifier (∃): it indicates that there exists at least one object for
which the statement within the quantifier is true
 Examples of Predicate Logic in AI:
1. Predicate Example: "IsHungry(x)"
Predicate Symbol: IsHungry
Argument: x (variable representing an object)
Meaning: These predicate asserts that a specific object represented by "x" is
hungry.
2. Universal Quantification: ∀x IsHuman(x) → IsMortal(x)
This statement uses the universal quantifier to claim that for all objects "x" in the
domain, if "x" is human, then "x" is mortal.
3. Existential Quantification: ∃x IsHungry(x)
This statement uses the existential quantifier to claim that there is at least one
object "x" in the domain that is hungry.
 Characteristics of Predicate Logic
* The Logical inference is allowed.

* More accurate knowledge representation of facts of the real world.

* Program designing is its application area.

* Better theoretical foundation.

 Applications of Predicate Logic:

* In predicate logic knowledge is represented in the form of facts & rules as in done
in prolog.
* Predicates help AI planners search for a sequence of actions to achieve a goal.
* Predicates allow machine learning algorithms to discover patterns and make
predictions
* Predicates are used in natural language processing for understanding the semantics
of sentences.
* Predicates allow expert systems to make informed decisions and solve problems by
using facts & rules.

7
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

1.5.3 Introduction to Programming Languages

A programming language includes syntax, semantics of programs and the
computation.
1) Procedural Language: The program specifies a computation by saying "how” it is
to be performed. FORTRAN, C, and Object-oriented languages fall under this general
approach.

2) Declarative Language: The program specifies a computation by giving the logical

properties of a correct answer. LISP, Prolog and Logic Data Language (LDL) are
called logic programming languages.

1.5.4 Logic Programming

 Logic program is a collection of logic statements which express facts and rules
about a problem domain.
 Logic programming specify a computation in terms of logical relations between
entities.
 Computation determines whether, a particular conclusion follows from those
logical statements.
 In logic programming, using the knowledge base (collection of facts and rules),
the machine can find answers to the given questions.

8
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Major programming language families include Prolog, LDL, LISP

 Characteristics of Logic program
1. Logic program is characterized by set of relations and
inferences. − program consists of a set of axioms and a goal statement.
2. Rules of inference determine whether the axioms are sufficient to ensure the truth
of the goal statement.
3. Execution of a logic program corresponds to the construction of a proof of the goal
statement from the axioms.
4. In logic programming, logic is used to represent knowledge and inference is used
to manipulate it.
1.5.5 PROLOG
Prolog as the name itself suggests, is the short form of PROgramming in LOGics

Definition: “It is a logical and declarative programming language designed for developing logic-
based AI applications”.

Developers can “set rules & facts” around a problem, and then Prolog’s interpreter will
use that information to automatically execute those plans to find solutions.

Prolog programs consist of set of three different Horn clauses (“: -") -

1. Facts − The fact is a predicate that is true

Syntax: Relation (object1, object2).
Example: Father (James, Robert)
2. Rules − Rules consists of a head and body separated by conditional clauses “: - “
Syntax: rule name (object1, object2): - fact (object1, object2).
Example: Parent (person1, person2): - Father (James, Robert).
3. Queries − To run a prolog program, we need some questions, and those questions
can be answered by the given facts and rules. Knowledge Base can be considered
like database, against which we can query.

9
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Query syntax: ?: - odd number (7).

Output: No.
Explanation: As our knowledge base does not contain the above fact, output
is No

 Prolog uses uppercase letters for variables and lowercase for constants.
 Commas separate literals in the body, and Period marks the end of a sentence
 Advantages:
1. Easy to build database. Doesn’t need a lot of programming effort.
2. Pattern matching is easy. Search is recursion based.
3. It has built in list handling. Makes it easier to play with any algorithm involving
lists.
 Disadvantages:
1. LISP logic programming language dominates over prolog with respect to I/O
features.
2. Sometimes input and output is not easy.
 Applications of Prolog:
* It plays a vital role in automation system.
* Intelligent Database Retrieval
* Natural Language Understanding
* Specification Language
* Machine Learning
* Robot Planning
* Problem Solving
1.6 Semantic Network Representation
* In Semantic networks, we can represent our knowledge in the form of graphical
networks.
* This network consists of nodes representing objects and arcs which describe the
relationship between those objects.
* Semantic networks can categorize the object in different forms and can also link those
objects.
* Semantic networks are easy to understand and can be easily extended.
* This representation consists of mainly two types of relations:
a. IS-A Relation (Inheritance)

10
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

b. Kind-of-Relation
Example: Statements are represented in the form of nodes and arcs.
Statements: a) Man is a human.
b) All humans has name.

In the above diagram, we have represented the different type of knowledge in the form
of nodes and arcs. Each object relates to another object by some relation.
 Advantages of Semantic network:
1. Semantic networks are a natural representation of knowledge.
2. Semantic networks convey meaning in a transparent manner.
3. These networks are simple and easily understandable.
 Drawbacks in Semantic representation:
1. Semantic networks take more computational time at runtime as we need to
traverse the complete network tree to answer some questions.
2. It is not possible to build a vast semantic network to model human-like memory
(Which has 1015 neurons and links) to store the information.
3. These types of representations are inadequate as they do not have any equivalent
quantifier, e.g., for all, for some, none, etc.
4. Difficulties in modelling sophisticated knowledge structures.
5. These networks are not intelligent and depend on the creator of the system.

1.7 Frame Representation

Frames in artificial intelligence are a structured knowledge representation technique. It
consists of a collection of slots and facets (fillers). A frame is also known as slot-filler
knowledge representation in artificial intelligence.
 Key Characteristics of Frames:
 Structure: Frames are organized hierarchically, with frames containing slots
(attributes) and fillers (values) representing specific information about objects or
concepts.

11
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Attributes and Values: Frames focus on modelling knowledge using attribute-

value pairs. Each frame has attributes (slots) associated with specific values
(fillers).
 Inheritance: Frames often support inheritance, allowing subframes to inherit
attributes and values from their parent frames.
 Use Cases: Frames are well-suited for structured representation in knowledge-
based systems, and expert systems. Including Natural language processing and
machine visions.
 Frame Representation example for a “person”

Slots(attribute) Filler(value)
Name John Smith
Age 35
Gender Male
Address 123 Main Street
Phone Number (555) 123-4567

 Advantages of frame representation:

1. It makes the programming easier by grouping the related data.
2. The frame representation is comparably flexible and used by many AI applications
3. It is very easy to add slots for new attribute and relations.
4. It is easy to include default data and to search for missing values.
5. Frame representation is easy to understand and visualize.
 Disadvantages of frame representation:
1. Inference mechanism cannot be smoothly proceeded by frame representation.
2. Frame representation has a much-generalized approach.

1.8 Frame Inheritance-Hierarchical Structures

Inheritance is an essential feature of frame-based system. In this hierarchical
arrangement, frames inherit properties and attributes from their parent frames. This means that
frames can automatically acquire and extend their knowledge, based on their position in
the frame hierarchy.
 Illustrating Frame Inheritance with an example:
In frame-based systems, frames are organized hierarchically.
 Consider a frame representing a "Car."

12
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 This frame might contain slots such as "Make," "Model," "Year," and "Color."
 Each of these slots has corresponding fillers, providing specific values.
 Additionally, the "Car" frame may be organized hierarchically within a larger
category called "Vehicle," inheriting certain properties and attributes.
 Advantages:
Inheritance simplifies knowledge representation in several ways:
 Reduces Redundancy: Inheritance eliminates the need to redundantly specify
common attributes and values for each specific frame.
 Maintains Consistency: Inheritance ensures that related frames share consistent
information.
 Facilitates Extensibility: New frames can be added to the hierarchy without the need
to specify all their attributes from scratch.
 Enables Classification: Inheritance allows frames to be classified based on their
position in the hierarchy.
 Supports Abstraction: Hierarchical frames can represent abstract and specific
concepts.
1.9 Representing Knowledge using Rules:
Rule based systems in AI express knowledge using a set of rules or production rules to
generate judgements or suggestions. Rules are expressed in the following form:
“If<condition> then <conclusion>”

1.9.1 Components of Rule-Based System

A typical rule-based system comprises several key components:

13
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

1. Rules are conditional statements that define the system’s behavior. “IF condition
THEN action.”
2. Knowledge Base: This is the repository where all the rules and facts are stored.
3. Inference Engine applies the rules to the knowledge base to derive conclusions or
make decisions.
4. Working Memory: This is a dynamic component that holds the current facts. It is
updated as the inference engine applies rules.
5. User Interface allows users to interact with the system, input data, and receive
outputs or recommendations.
1.9.2 Rule-based Deduction System
A rule-based deduction system is a computer system that uses rules & reasoning to solve
problems in a specific domain. The rules are often written as if-then statements, such as
"If X happens, then do Y”.
Here’s how a rule-based system works:
1. Create rules and facts: A developer creates a list of rules and facts for the system.
2. Analyse data: The system processes the input data against the rules.
3. Perform actions: The system follows the rules and performs any programmed
functions.
 Advantages
 Accuracy: Rule-based systems operate by cause and effect, and only within their
rule set.
 Ease of use: Rule-based systems require only small amounts of simple data to
perform tasks and repetitive processes.
 Speed: With the proper training, rule-based systems can make informed decisions
quickly and efficiently.
 Disadvantages:

14
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Lack of Learning Capability: Rule-based systems are exact and do not have
learning capabilities. Including too many rules can slow down a system and
introduce complexity.
 Scalability: Altering existing rules or incorporating new rules can introduce
time-consuming and expensive complications.
 High Maintenance Costs: To keep the rules accurate and up to date, rule-based
systems need continual maintenance.
 Applications of Rule-based System
A classic example of a rule-based system is the domain-specific expert system that
uses rules to make deductions or choices.
 Help a doctor choose the correct diagnosis based on a cluster of symptoms.
 Select tactical moves to play a game.
2. Constraint Propagation in AI
Artificial Intelligence (AI) encompasses a variety of methods and techniques to solve
complex problems efficiently. One such technique is constraint propagation, which plays a
crucial role in areas like scheduling, planning, and resource allocation. A method of
inference that assigns values to variables characterizing a problem in such a way that some
conditions (called constraints) are satisfied. This process can result in more domain
reductions.

Definition: Constraint propagation is a process in artificial intelligence (AI) that uses

“constraints to reduce the possible values of variables and find new constraints”.

Key Concepts
1. Variables: Elements that need to be assigned values.
2. Domains: Possible values that can be assigned to the variables.
3. Constraints: Rules that define permissible combinations of values for the variables.
How Constraint Propagation Works

 Constraint propagation works by iteratively narrowing down the domains of

variables based on the constraints.
 This process continues until no more values can be eliminated from any domain.
The primary goal is to reduce the search space and make it easier to find a solution.

Steps in Constraint Propagation

15
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

1. Initialization: Start with the initial domains of all variables.

2. Propagation: Apply constraints to reduce the domains of variables.
3. Iteration: Repeat the propagation step until a stable state is reached, where no
further reduction is possible.

Example: Consider a simple CSP with two variables, X and Y, each with domains {1, 2,
3}, and a constraint X ≠ Y.
Constraint propagation will iteratively reduce the domains as follows:
 If X is assigned 1, then Y cannot be 1, so Y’s domain becomes {2, 3}.
 If Y is then assigned 2, X cannot be 2, so X’s domain is reduced to {1, 3}.
 This process continues until a stable state is reached.
3. Reasoning in AI
Reasoning in Artificial Intelligence refers to the “process by which AI systems analyse
information, make inferences, and draw conclusions to solve problems or make decisions”.
Types of Reasoning in AI
1) Probabilistic Reasoning in AI: Probabilistic reasoning involves dealing with
uncertainty and making decisions based on probabilities. AI systems use statistical
models to assess the likelihood of different outcomes and make informed choices.
2) Default Reasoning in AI: It is a type of non-monotonic reasoning where conclusions
are drawn based on default assumptions unless explicitly contradicted.
3) Statistical Reasoning in AI: statistical reasoning involves the use of statistical
methods to analyse data, identify patterns, and make predictions.

3.1 Probabilistic Reasoning with uncertainty in Artificial intelligence

A) Uncertainty:
 Till now, knowledge representation using first-order logic and propositional logic
gives certainty, which means we were sure about the predicates.
 With this knowledge representation, we might write A→B, which means if A is true
then B is true.
 But consider a situation where we are not sure about whether A is true or not then we
cannot express this statement, this situation is called uncertainty.

16
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 So, to represent uncertain knowledge, where we are not sure about the predicates, we
need uncertain reasoning or probabilistic reasoning.
B) Causes of uncertainty:
Following are some leading causes of uncertainty to occur in the real world.
1. Information occurred from unreliable sources.
2. Experimental Errors
3. Equipment fault
4. Temperature variation
5. Climate change.
C) Probabilistic Reasoning:
Probabilistic Reasoning is a way of knowledge representation where we apply the
concept of probability with logic to indicate the uncertainty in knowledge.

In probabilistic reasoning, there are two ways to solve problems with uncertain
knowledge:
o Bayes' rule
o Bayesian Statistics
3.2 Review of Probability
Let's understand some common terms:
1. Probability: Probability can be defined “as a chance that an uncertain event will occur.”
It is the numerical measure of the likelihood that an event will occur. The value of
probability always remains between 0 and 1 that represent ideal uncertainties.
* 0 ≤ P(A) ≤ 1, where P(A) is the probability of an event A.
* P(A) = 0, indicates total uncertainty in an event A.
* P(A) =1, indicates total certainty in an event A.

We can find the probability of an uncertain event by using the below formula.

2. Event: Each possible outcome of a variable is called an event.

3. Sample space: The collection of all possible events is called sample space.
4. Random variables: Random variables are used to represent the events and objects in
the real world.

17
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

5. Prior probability: The prior probability of an event is probability computed before

observing new information.
6. Posterior Probability: The probability that is calculated after all evidence or
information has considered. It is a combination of prior probability and new
information.
7. Conditional probability is a “probability of occurring an event when another event
has already happened”.
Let’s suppose, we want to calculate the event A when event B has already occurred,
* The probability of A under the conditions of B, it can be written as:

Where, P(A⋀B) = Joint probability of A and B

P(B) = Marginal probability of B.
* If the probability of A is given and we need to find the probability of B, then it
will be given

It can be explained by using the below Venn diagram, where B has occurred event, so
sample space will be reduced to set B, and now we can only calculate event A when
event B has already occurred by dividing the probability of P(A⋀B) by P (B).

3.3 Bayes Theorem

In AI, Bayes' Theorem is used in probabilistic reasoning, machine learning, and data
science to model and handle uncertainty. Bayes theorem can be derived using product rule
and conditional probability of event A with known event B:

P (A ⋀ B) = P(A|B) P(B)
OR
P (A ⋀ B) = P(B|A) P(A)

Equating right hand side of both the equations, we will get:

18
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

The above equation (a) is called as Bayes' rule or Bayes' theorem. This equation is basic
of most modern AI systems for probabilistic inference.

* P(A|B) is known as posterior, which we need to calculate, and it will be read as

Probability of hypothesis A when we have occurred an evidence B.
* P(B|A) is called the likelihood, in which we consider that hypothesis is true, then
we calculate the probability of evidence.
* P(A) is called the prior probability, probability of hypothesis before considering the
evidence.
* P(B) is called marginal probability, pure probability of an evidence.
 Application of Bayes' theorem in Artificial intelligence:

* It is used to calculate the next step of the robot when the already executed step is
given.
* Bayes' theorem is helpful in weather forecasting.
* It can solve the Monty Hall problem.
* Applications include spam email classification, medical diagnosis, natural language
processing, and Bayesian networks.

3.4 Bayesian Networks (aka Belief Networks)

Bayesian Networks, also known as Bayes Nets, Belief Nets, Causal Nets, and
Probability Nets are “space-efficient data structure for computing any value in the full joint
probability distribution of the set of random variables”.

Bayesian Network can be used for building models from data and expert’s opinions, and
it consists of two parts:

 Directed Acyclic Graph

 Table of conditional probabilities.
A Bayesian network graph is made up of nodes and Arcs (directed links), where:

 Each node corresponds to the random variables, and a variable can be continuous
or discrete.

19
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Arc or directed arrows represent the causal relationship or conditional probabilities

between random variables. These directed links or arrows connect the pair of nodes
in the graph.

Properties:
1. Can be used to reason:
o Forward (top-down) from causes to effects -- predictive reasoning (causal
reasoning)
o Backward (bottom-up) from effects to causes -- diagnostic reasoning
2. Captures both qualitative and quantitative relationships between variables

3.5 Dempster Shafer Theory

Dempster-Shafer Theory is a “mathematical framework for reasoning about uncertainty”
which was proposed by Arthur P. Dempster in 1967 and Glenn Shafer in the 1960s. This
theory was released because of the following reason: -

* Bayesian theory is only concerned about single evidence.

* Bayesian probability cannot describe ignorance.

Definition: Dempster-Shafer theory (DST), also known as belief function theory or

evidence theory, is a framework for representing, quantifying, and managing uncertainty
in artificial intelligence.

Dempster Shafer Theory stands as a valuable tool in the field of artificial intelligence to
solve problems, handle uncertainty effectively and make more accurate decisions.

3.5.1 The uncertainty in this model is given by: -

20
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

1. Conflict of evidence: In DST, uncertainty arises from conflicting evidence or

incomplete information. It Considers all possible outcomes.

2. Combination Rule: Dempster’s rule of combination, combine belief functions from

different sources.

3. Mass functions: In particular, a mass function must assign a mass of zero to the
empty set (corresponding to no information), and a mass of one to the entire set
(corresponding to complete certainty).

4. Belief and plausibility: Given a mass function, we can calculate two measures of
uncertainty. They are:

a) Belief: The belief of a set A is the sum of the masses of all the focal elements
which include A. Belief represents the lower bound of uncertainty (how much
we believe A). A belief function assigns a degree of belief (or plausibility) to
each possible outcome.
b) Plausibility is the sum of the masses of all the focal elements which intersect
A. Plausibility represents the upper bound of uncertainty (how much we can
accept A). Plausibility (denoted by Pl) is thus related to Bel by:
Pl(p) = 1 − Bel(~p). It also ranges from 0 to 1

Dempster–Shafer Theory in AI, can analyse the evidence, assign masses to subsets of
possible conclusions, and calculate beliefs and plausibility where, belief<=
plausibility

3.5.2 Characteristics of Dempster Shafer Theory

Dempster Shafer Theory in artificial intelligence (AI) exhibits several notable
characteristics:

1. Uncertainty Representation: it provides a way to represent and reason

incomplete evidence.
2. Conflict of Evidence: The DST allows for the combination of multiple sources
of evidence of evidence.
3. Handling Ignorance: Dempster Shafer Theory, ignorance is gradually
diminished through the accumulation of additional evidence.
4. Decision-Making Ability: By deriving measures such as belief, probability and
plausibility from the combined belief function it helps in decision making.

21
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

3.5.3 Advantages & Disadvantages

Advantages
* As we add more information, the uncertainty interval reduces.
* DST has a much lower level of ignorance.
* Enhances the robustness of decision-making processes in AI systems.
* Handling of incomplete or conflicting information encountered in artificial
intelligence.
Disadvantages
* In this, computation complexity is high, as we must deal with 2n sets
* The process of combining evidence necessitates careful modelling and
calibration.
* The interpretation of belief and plausibility values in DST may possess
subjectivity, biases.

*************

Review Questions Unit-III

PART-A (Two Marks)

1. Define knowledge representation and list its properties. Refer 1.1

2. Write the merits and demerits of Logical Representation. Refer 1.5.1
3. What is semantic nets representation. Give the diagrammatic representation with an
example.
Refer 1.6
4. Write the Frame representation for “Peter is an engineer as a profession, and his age is 25,
he lives in city London, and the country is England” Refer 1.7
5. Define predicate logic. Name the components of predicate logic. Refer 1.5.1(A)
6. Compare declarative and procedural representation of knowledge using rules. Refer 1.9
7. Define Logic programming in AI. Refer 1.5.4
8. What is reasoning in AI? Write the types of reasoning in AI. Refer 3
9. Define probability. What is the formula for probability of occurrence. Refer 3.2
10. State Dempster Shafer Theory. Refer 3.5

PART-B (Ten Marks)

11. Explain knowledge representation in AI? Refer1.1,1.2

12. Illustrate knowledge representation using Predicate Logic. Refer 1.5.1(A)

13. Explain Semantic Network Representation. Refer 1.6

22
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
 Department of CSE (AI & ML) II Year B.Tech-I Semester

14. What are Frames in AI? Demonstrate with example. Refer 1.7
15. Discuss in detail about Rule Based Systems. Refer 1.9
16. How constraint propagation works in AI. Illustrate with example. Refer 2
17. Illustrate Frame Inheritance with example. Refer 1.8
18. Explain Probabilistic Reasoning under uncertainty. Refer 3.1
19. Describe Bayes’ Theorem and its application in AI. Refer 3.3
20. Elaborate Dempster Shafer Theory. Refer 3.5

***********

LECTURE PLAN
UNIT IV
S.No. Topic No. of Proposed Actual Pertaining Taxonomy Mode of
Periods lecture lecture CO(s) Level Delivery

Period Period
1
First order 1 CO4 L2 MD1,
logic MD5
2 Inference in MD1,
first order 1 CO4 L2 MD5
logic
3 Propostional 1 MD1,
vs First order CO4 L2 MD5
inference
4 Unification
& Lifts 1 CO4 L1 MD1,
MD5
5 Forward 1 CO4 L1 MD1,
chaining MD5

23
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
 Department of CSE (AI & ML) II Year B.Tech-I Semester

6 Backward 1 CO4 L1 MD1,

chaining MD5
7 Resolution
1 CO4 L1 MD1,
MD5
8 Learning from 1
observation CO4 L2 MD1,
MD5

9 Inductive 1
learning CO4 L2 MD1,
MD5
10 Decision Tree 1 CO4 L2 MD1,
MD5
11 Explanation 1 CO4 L2 MD1,
based learning MD5
12 Statistical 1 CO4 L2 MD1,
learning MD5
methods
13 Reinforcement 1 CO4 L2 MD1,
Learning MD5

UNIT - IV
Logic concepts: First order logic. Inference in first order logic, propositional vs. first order
inference, unification & lifts forward chaining, Backward chaining, Resolution, learning from
observation Inductive learning, Decision trees, Explanation based learning, Statistical
Learning methods, Reinforcement Learning.

1. Introduction
Logical AI involves representing knowledge of an agent's world, its goals and the
current situation by sentences in logic. Logical AI deal with using knowledge base
and rule sets to make intelligent decisions.

24
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

An Inference Engine is a component of the expert system that applies logical rules to the
knowledge base to deduce new information. It interprets and evaluates the facts in the
knowledge base to provide an answer. A knowledgebase is a structured collection of facts about
the system’s domain.
1.1 First-Order logic

First Order Logic (FOL) can be defined as “a collection of objects, their attributes, and relations
among them to represent knowledge”. It's also known as Predicate Logic.
 Objects: A, B, people, numbers, colors, squares, pits, wars, theories, wumpus, ......
 Relations: It can be unary relation such as: red, round, is adjacent, or n-any relation
such as: the sister of, brother of, has colour, comes between
 Function: Father of, best friend, third inning of, end of, ......
 First-order logic also has two main parts as a natural language:
o Syntax
o Semantics

Syntax of First-Order logic: In first-order logic, the syntax of FOL determines which set of
symbols represents a logical expression.

The basic elements of FOL syntax are as follows:

Constant 1, 2, A, John, Mumbai, cat,
Variables x, y, z, a, b,
Predicates Brother, Father, >,
Functions sqrt,
Connectives ∧, v, ¬, ⇒, ⇔

25
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
 Department of CSE (AI & ML) II Year B.Tech-I Semester

Equality ==
Quantifier ∀, ∃

a) Atomic sentences: Predicate can be used to represent atomic sentences (term1, term2,
term3......, term n).

Ravi and Ajay are brothers: => Brothers (Ravi, Ajay).

b) Complex Sentences: Connectives are used to join atomic sentences to form complex
sentences.
c) Subject: The major component of the sentence is the subject.
d) Predicate: A predicate is a relationship that ties two atoms together in a sentence.
Consider the following statement: "x is an integer."

e) Quantifiers: These are the symbols that allow you to determine or identify the variable's
range and scope in a logical expression. There are two different kinds of quantifiers:
Universal Quantifier (for all, everyone, everything): Universal Quantifier used to express
that a statement is true for all objects in the domain.
Statement: All man drink coffee
Logic representation: ∀xman(x)→drink(coffee).
It will be read as: There are all x where x is a man who drink coffee.
Existential Quantifier, (for some, at least one): Existential quantifiers are a sort of quantifier
that expresses that a statement is true for at least one instance of something within its scope.
Statement: Some boys are intelligent.
Logic representation: ∃x: boys(x)∧intelligent(x)
It will be read as: There are some x where x is a boy who is intelligent.
1.2 Inference in First Order Logic

In artificial intelligence, we need intelligent computers which can create new logic from
old logic or by evidence, so generating the conclusions from evidence and facts is termed
as Inference. “Inference in First-Order Logic is used to deduce new facts or sentences from existing
sentences”.

A. Substitution:

26
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Substitution is a fundamental operation performed on terms and formulas. It occurs in all

inference systems in first-order logic. The substitution is complex in the presence of
quantifiers in FOL. If we write F[a/x], so it refers to substitute a constant "a" in place of
variable "x".
B. FOL Inference rules for quantifiers:

There are some Inference rules that can be applied to sentences with quantifiers to
obtain sentences without quantifiers. These rules will lead us to make the conversion.

1. Universal Generalization: This rule can be used if we want to show that every
element has a similar property. It can be represented as:

Ex: P(c): "A byte contains 8 bits",

∀ x P(x): "All bytes contain 8 bits.", it will also be true.
2. Universal Instantiation: The UI rule state that we can infer any sentence P(c) by
substituting a ground term c (a constant within domain x) from ∀ x P(x) for any object
in the universe of discourse. It can be represented as:

Ex: IF "Every person like ice-cream"=> ∀x P(x) so, we can infer that,
“John likes ice-cream" => P(c)
3. Existential Instantiation is also known as Existential Elimination. This rule states
that one can infer P(c) from the formula given in the form of ∃x P(x) for a new
constant symbol c. It's written like this:

Ex: From the given sentence: ∃x Crown(x) ∧ OnHead(x, John),So we can

infer: Crown(K) ∧ OnHead(K, John),
The above used K is a constant symbol, which is called Skolem constant. The
Existential instantiation is a special case of Skolemization process.
4. Existential introduction is also known as an Existential Generalization. This rule
states that if there is some element c in the universe of discourse which has a
property P, then we can infer that there exists something in the universe which

has the property P. It's written like this:

27
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Ex: Let's say that "Priyanka got good marks in English."

Therefore, “someone got good marks in English”

C. Generalized Modus Ponens Rule:

For the inference process in FOL, we have a single inference rule which is called
Generalized Modus Ponens. It is lifted version of Modus ponens. Generalized Modus
Ponens can be summarized as, " P implies Q and P is asserted to be true, therefore Q must be
True."
According to Modus Ponens, for atomic sentences pi, pi', q. Where there is a substitution
θ such that SUBST (θ, pi',) = SUBST (θ, pi), it can be represented as:

1.3 Propositional Logic Vs. First Order Logic

In knowledge representation, two fundamental forms of logic are used:

a.Propositional Logic b. First Order Logic

A. Propositional logic: Propositional logic or sentential logic, deals with propositions that can
be either true or false and uses logical connectives to form complex expressions. However,
propositional logic is limited because it cannot represent relationships between objects
or quantify over them.

The basic components of propositional logic include:

 Propositions: Basic statements that are either true or false.

 Logical Connectives: Operators such as AND (∧), OR (∨), NOT (¬), IMPLIES (→)

and BICONDITIONAL (↔) used to combine propositions.

 Truth Values: Each proposition has a truth value of either true (T) or false (F).

28
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Example:
 P: “It is raining.”
 Q: “The ground is wet.”

Using logical connectives, we can form complex expressions like P→Q (If it is
raining, then the ground is wet).
 Applications:
 Digital Circuit Design: Representing and analysing the behaviour of logic gates and
circuits where each component can be represented as a true/false variable, and the
overall circuit can be analysed using logical connectives.
 Expert Systems: Encoding simple rules and facts for decision-making systems.
 Truth Tables: Evaluating the truth values of logical expressions based on various
combinations of input values.
B. First-order logic: First-order logic, or predicate logic, extends propositional logic by
introducing predicates and quantifiers, allowing us to express more complex statements. First-
order logic cannot represent higher-order concepts, such as statements about other
statements or sets of objects.

The basic components of first-order logic include:

* Constants: Specific objects in the domain (e.g., Alice, Bob).

* Variables: Symbols that can represent any object in the domain (e.g., x, y).

* Predicates: Functions that map objects to truth values (e.g., Likes (Alice, Ice
Cream)).
* Quantifiers: Symbols that indicate the scope of a statement (e.g., ∀ (for all), ∃

(exists)).
* Logical Connectives: Same as in propositional logic.

29
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Applications:
* Ontology Modelling: Representing knowledge about categories, properties, and
relationships between concepts in a domain.
* Semantic Web: Encoding information about web resources and their relationships
to enable intelligent searching and data integration.
* Automated Reasoning: Developing systems that can reason about knowledge,
make inferences, and answer queries based on a set of axioms and rules.
* Natural Language Processing: Understanding and generating human language by
modelling the relationships and properties of words and sentences.
C. Key Differences Summarized:

Feature Propositional Logic First-Order Logic

Basic Unit Propositions Predicates, constants, variables
Limited to true/false statements. Not Expressive, like natural language it
Expressiveness able to represent natural language can represent objects relations and
statements functions.
Quantifiers None Universal (∀) and Existential (∃)
Combines propositions using logical
Syntax Uses predicates and quantifiers
connectives

Semantics Truth tables Interpretation over a domain

Unification, forward chaining,
Inference
DPLL, GSAT Fast in practice backward chaining, Prolog, theorem
Algorithms
proving.
Simple problems (e.g., circuit design, Complex problems (e.g., AI reasoning,
Use Cases
rule-based systems) ontology modeling)

Example P→Q

1.4 Unification

30
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Unification is a process of making two different logical atomic expressions identical by

finding a substitution. “It takes two literals as input and makes them identical using
substitution”. Unification is a key component of all first-order Inference algorithms.
Let Ψ1 and Ψ2 be two atomic sentences and 𝜎 be a unifier such that, Ψ1𝜎 = Ψ2𝜎, then it
can be expressed as UNIFY (Ψ1, Ψ2).
UNIFY (p, q) = θ where SUBST (θ, p) = SUBST (θ, q)
θ is our Most General Unifier (MGU) value (if one exists).
P (x, y) ......... (i)
P (a, f(z)) ......... (ii)
Substitute x with a, and y with f(z) in the first expression, and it will be represented
as a/x and f(z)/y.

With both the substitutions, the first expression will be identical to the second expression
and the substitution set will be: [a/x, f(z)/y].

Following are some basic conditions for unification:

o Predicate symbol must be same.
o Number of Arguments in both expressions must be identical.
o Unification will fail if there are two similar variables present in the same expression.

1.5 Lifting
Lifting is a concept that extends the utility of unification, enabling more flexible and
abstract handling of knowledge representations in AI. Lifted Inference rule require
finding substitutions that make different logical expressions look identical (same). That
is unification.
It is often associated with answer set programming (ASP), a logic programming
paradigm that serves as a foundation for AI knowledge representation and reasoning.
1.6 Resolution:
Resolution is used, if there are various statements are given, and we need to prove a
conclusion of those statements. Unification is a key concept in proofs by resolutions.
Resolution is a single inference rule which can efficiently operate on the conjunctive normal form
or clausal form. Every sentence of first-order logic can be converted into an inferentially
equivalent CNF sentence.
* Clause: Disjunction of literals (an atomic sentence) is called a clause. It is also
known as a unit clause.

31
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

* Conjunctive Normal Form: A sentence represented as a conjunction of clauses is

said to be conjunctive normal form (CNF).

This rule is also called the binary resolution rule because it only resolves exactly two
literals.
 Steps for Resolution:
1. Conversion of facts into first-order logic.
2. Convert FOL statements into CNF
3. Negate the statement which needs to prove (proof by contradiction)
4. Draw resolution graph (unification).
 The resolution inference rule:
The resolution rule for first-order logic is simply a lifted version of the propositional
rule. Resolution can resolve two clauses if they contain complementary literals, which
are assumed to be standardized apart so that they share no variables.

Where, li and mj are complementary literals.

1.7 Forward Chaining and Backward Chaining in AI

In First-Order Logic, inference is used to derive new facts or sentences from existing
ones. To recommend a solution, the Inference Engine (part of expert system) uses the
following strategies:
a. Forward Chaining b. Backward Chaining
A. Forward Chaining:
It is a strategy of an expert system to answer the question, “What can happen next?”
Here, the Inference Engine follows the chain of conditions and derivations and finally
deduces the outcome. Forward chaining is a form of reasoning that starts with simple facts
in the knowledge base and applies inference rules in the forward direction to extract more data
until a goal is reached.
For example, prediction of share market status as an effect of changes in interest rates.

32
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Properties of Forward-Chaining:
• It is a process of making a conclusion based on known facts or data, by starting from
the initial state and reaches the goal state.
• It is a down-up approach, as it moves from bottom to top.

• Also called as data driven as we reach to the goal using available data.

• It applies the Breadth-First Strategy.

• Its goal is to get the conclusion.

• It applies inference rules (Modus Ponens) in the forward direction to extract more data
until a goal is reached.
• Slow as it must use all the rules.

• Commonly used in the Expert Systems, such as CLIPS, business, and production rule

systems.

B. Backward Chaining:
 With this strategy, an expert system finds out the answer to the question, “Why
this happened?”. Based on what has already happened, the Inference Engine tries to
find out which conditions could have happened in the past for this result.
 This strategy is followed for finding out cause or reason. This algorithm works
backward from the goal, chaining through rules to find known facts that support the proof.
For example, diagnosis of blood cancer in humans.

 Properties of backward chaining:

• In backward chaining, the goal is broken into sub-goal or sub-goals to prove the
facts true.
• It is known as a top-down approach.
• It is called a goal-driven approach, as a list of goals decides which rules are selected
and used.

33
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

• It operates in backward direction i.e. it works from goal to reach initial state.
• The backward-chaining method mostly used a depth-first search strategy for proof.
• Its goal is to get the possible facts, or the required data based on Modus Ponens
inference rule.
• Fast as it must use only a few rules.
• Backward-chaining algorithm is used in game theory, automated theorem proving tools,
inference engines, proof assistants, and various AI applications.
2. Learning
Learning is essential for unknown environments. It modifies the agent's decision
mechanisms to improve performance.
* “Learning is constructing or modifying representations of what is being experienced”.
(Michalski, 1986)
* “A computer program learns if it improves its performance at some tasks through
experience” (Mitchell, 1997).
 Machine Learning:
In the field of artificial intelligence known as machine learning, algorithms are
developed that can learn patterns and make judgments or predictions without being
explicitly programmed.

 Categorization of Machine Learning Algorithm:

34
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 The two main methods used in machine learning are:

2.1 Observational learning

35
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Observational learning is an important concept in artificial intelligence (AI). AI systems

can learn by observing the behaviour of humans or other AI systems. It is a process where
humans or animals learn from each other’s actions, gestures, and behavior.
There are four key elements that are involved in observational learning:
1. Attention – the individual needs to pay attention to the behavior that is being
demonstrated.
2. Retention – the individual needs to remember or retain the information or behavior
they have observed.
3. Reproduction – the individual needs to be able to reproduce the behavior that they
have observed.
4. Motivation – the individual needs to be motivated to reproduce the behavior that
they have observed.
For example, an AI system can learn to play a game by observing the behavior of a human player.
The AI system can observe the human's movements, decision-making process, and
strategies, and then use that information to improve its own performance in the game.
2.2 Inductive Learning Algorithm (ILA)
Inductive Learning Algorithm (ILA) is an iterative and inductive machine learning algorithm
that is used for generating a set of classification rules, which produces rules of the form “IF-
THEN”, for a set of examples, producing rules at each iteration and appending to the set
of rules.
 Advantages:
* Inductive learning models are flexible and adaptive, they are well suited for
handling difficult, complex, and dynamic information.
* Finding hidden patterns and relationships in data.
* They can efficiently handle huge volumes of data.
* Inductive learning models may learn from examples without explicit programming
where the rules are ambiguous
 Disadvantages:
* May overfit to data
* computationally costly
* Limited interpretability
* If the data is inaccurate or inadequate, the model may not perform effectively.

 Applications: Inductive learning algorithms are used in a variety of applications,

including credit risk assessment, disease diagnosis, face recognition, and autonomous driving.

36
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Illustration with an Example:

Example Place type weather location decision
No.
1. hilly winter kullu Yes
2. mountain windy Mumbai No
3. mountain windy Shimla Yes
4. beach windy Mumbai No
5. beach warm goa Yes
6. beach windy goa No
7. beach warm Shimla Yes
 Subset – 1

Example Place type weather location decision

No.
1. hilly winter kullu Yes
2. mountain windy Shimla Yes
3. beach warm goa Yes
4. beach warm Shimla Yes
 Subset – 2

Example Place type weather location decision

No.
5. mountain windy Mumbai No
6. beach windy Mumbai No
7. beach windy goa No
 Finally, we get the Rule Set:

* Rule 1: IF the weather is warm THEN the decision is yes.

* Rule 2: IF the place type is hilly THEN the decision is yes.
* Rule 3: IF the location is Shimla, THEN the decision is yes.

* Rule 4: IF the location is Mumbai, THEN the decision is no.

* Rule 5: IF the place type is beach AND the weather is windy THEN the decision is
No

2.3 Decision Tree Learning:

A Decision tree is a tree-like structure that represents a set of decisions and their possible
consequences. Each node in the tree represents a decision, and each branch represents an

37
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

outcome of that decision. The leaves of the tree represent the final decisions or
predictions. Decision trees are used in various fields such as machine learning, data mining,
and statistics.

 Structure of a Decision Trees:

* Root Node: The decision tree’s starting node, which represents the complete dataset.
* Leaf Node: Leaf nodes are the final output node, and the tree cannot be divided
further after getting a leaf node.
* Branches/Sub Tree: A tree formed by splitting the tree.
* Splitting: Splitting is the process of dividing the decision node/root node into sub-
nodes according to the given conditions.
* Splitting Criteria: Metrics like information gain, entropy, or the Gini Index are used
to calculate the optimal split.
* Decision Rules: Rules that govern the splitting of data at each branch node.
* Attribute Selection: The process of choosing the most informative attribute for each
split.
* Pruning: Pruning is the process of removing the unwanted branches from the tree.
 How does the Decision Tree algorithm Work?
* Step-1: Begin the tree with the root node, says S, which contains the complete dataset
* Step-2: Find the best attribute in the dataset using Attribute Selection Measure (ASM).
* Step-3: Divide the S into subsets that contains possible values for the best attributes
* Step-4: Generate the decision tree node, which contains the best attribute.
* Step-5: Recursively Continue this process until a stage is reached where you cannot
further classify the nodes and called the final node as a leaf node.
 Illustration with an example:
This decision tree classifies Saturday mornings according to whether they are suitable
for playing tennis.

38
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

For example, the instance (Outlook = Sunny, Temperature = Hot, Humidity = High,
Wind = Strong) would be sorted down the leftmost branch of this decision tree and
would therefore be classified as a negative instance i.e., the tree predicts that Play
Tennis = no.

2.4 Explanation-Based Learning

Explanation-based learning in artificial intelligence is a problem-solving method that
involves agent learning by analysing specific situations and connecting them to previously
acquired information. EBL algorithms incorporate logical reasoning and domain knowledge
to make predictions and identify patterns.

 Key Characteristics of EBL:

i. Use of Domain Knowledge: This knowledge helps the system to generalize the
learned concept to new, similar situations.
ii. Focused Learning: EBL focuses on understanding the essential features of an
example that are necessary to achieve a goal or solve a problem.
iii. Efficiency: Since EBL can learn from a single example, it is computationally efficient
compared to other learning methods that require large datasets for training.
 How Explanation-Based Learning Works?

39
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Explanation-Based Learning follows a systematic process that involves the following

steps:
1. Input Example: The learning process begins with a single example that the
system needs to learn from. This example is typically a positive instance of a
concept that the system needs to understand.
2. Domain Knowledge: The system uses domain knowledge, which includes rules,
concepts, and relationships relevant to the problem domain. This knowledge is
crucial for explaining why the example is valid.
3. Explanation Generation: The system generates an explanation for why the
example satisfies the concept. This involves identifying the relevant features and
their relationships that make the example a valid instance.
4. Generalization: Once the explanation is generated, the system generalizes it to
form a broader concept that can apply to other similar examples. This
generalization is typically in the form of a rule or a set of rules that describe the
concept.
5. Learning Outcome: The outcome of EBL is a generalized rule or concept that can
be applied to new situations. The system can now use this rule to identify or solve
similar problems in the future.
 Advantages of Explanation-Based Learning
 Efficiency in Learning: EBL can learn effectively from a single example, making it
efficient in situations where data is scarce or expensive to obtain.
 Understanding and Generalization: EBL focuses on understanding the rationale
behind examples, leading to more robust generalizations that can be applied to a
wide range of situations.
 Interpretable Models: The rules or concepts learned through EBL are often more
interpretable than those learned through other methods, making it easier to
understand and trust the system’s decisions.
 Challenges and Limitations
 Dependency on Domain Knowledge: EBL relies heavily on accurate and
comprehensive domain knowledge. If the domain knowledge is incomplete or
incorrect, the system may generate flawed explanations and generalizations.
 Limited to Well-Defined Problems: EBL is most effective in well-defined problem
domains where the rules and relationships are clear. It may struggle in more complex
or ambiguous domains.

40
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Complexity of Explanation Generation: Generating explanations can be

computationally intensive, especially in domains with complex relationships and many
features.
 Applications
Explanation-based learning in the artificial intelligence approach has a wide range
of applications, from medical diagnosis to fraud detection.
2.5 Statistical Machine Learning
Statistical Machine Learning involves using statistical techniques to develop models that can
learn from data and make predictions or decisions.

Popular statistical machine learning techniques are:

1. Linear Regression: It is also seen as a supervised learning algorithm that captures the
connection between a dependent variable and independent variables.
Statistics assist in estimating coefficients, conducting hypothesis tests, and evaluating
the significance of the relationships, providing valuable insights and a deeper
understanding of the data.
2. Logistic Regression: Logistic regression is a statistical classification algorithm that
estimates the probability of categorical outcomes based on independent variables. By applying
a logistic function, it predicts the occurrence of a particular class.
3. Decision-Tree: Decision trees are versatile algorithms that use statistics to split data
based on features, creating a tree-like structure for classification or regression. They are
intuitive, interpretable, and handle categorical and numerical data. Statistics-based
measurements, such as Gini impurity or information gain, are often incorporated to
guide the splits throughout the tree construction process.
4. Random forest: Random Forest is an ensemble learning method that improves prediction
accuracy by combining multiple decision trees. It employs sampling to randomly select
subsets of features and data for building the trees. The predictions of these individual
trees are then aggregated to make the final prediction.
5. SVM: SVM is a powerful algorithm that can be used for classification and regression
tasks. It uses statistical principles to create a boundary between different groups of data
points, making it easier to tell them apart. By optimizing this boundary, SVM reduces
the chances of making mistakes and improves overall accuracy.

41
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

6. K-Nearest neighbour KNN: KNN is a simple, yet effective algorithm used for
classifying data points based on the majority vote of their nearest neighbors. It is suitable for
both classification and regression problems and does not require training.
In KNN, statistical measures are utilized to determine the proximity between data
points, helping to identify the nearest neighbors. The majority vote of the nearest
neighbors is then used to classify or predict the target variable.
2.6 Reinforcement Learning
Reinforcement Learning (RL) is a branch of machine learning focused on making decisions to
maximize cumulative rewards in each situation. RL involves learning through experience. In
RL, an agent learns to achieve a goal in an uncertain, potentially complex environment by
performing actions and receiving feedback through rewards or penalties.

 Key Concepts of Reinforcement Learning:

* Agent: The learner or decision-maker.

* Environment: Everything the agent interacts with.

* State: A specific situation in which the agent finds itself.

* Action: All possible moves the agent can make.

* Reward: Feedback from the environment based on the action taken.

 How Reinforcement Learning Works

RL operates on the principle of learning optimal behavior through trial and error. The
agent takes actions within the environment, receives rewards or penalties, and adjusts
its behavior to maximize the cumulative reward.

42
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

This learning process is characterized by the following elements:

* Policy: A strategy used by the agent to determine the next action based on the
current state.
* Reward Function: A function that provides a scalar feedback signal based on the
state and action.
* Value Function: A function that estimates the expected cumulative reward from a
given state.
* Model of the Environment: A representation of the environment that helps in
planning by predicting future states and rewards.
 Main points in Reinforcement learning –
* Input: The input should be an initial state from which the model will start
* Output: There are many possible outputs as there are a variety of solutions to a
particular problem.
* Training: The training is based upon the input; The model will return a state, and
the user will decide to reward or punish the model based on its output.
* The model keeps continues to learn.
* The best solution is decided based on the maximum reward.
 Advantages:
1. Reinforcement learning can be used to solve very complex problems that cannot be
solved by conventional techniques.
2. The model can correct the errors that occurred during the training process.
3. In RL, training data is obtained via the direct interaction of the agent with the
environment
4. This is useful in real-world applications where the environment may change over
time or is uncertain.
5. Reinforcement learning can be used to solve a wide range of problems, including
those that involve decision making, control, and optimization.
6. Reinforcement learning is a flexible approach that can be combined with other
machine learning techniques, such as deep learning, to improve performance.

43
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Disadvantages:
1. Reinforcement learning is not preferable to use for solving simple problems.
2. Reinforcement learning needs a lot of data and a lot of computation.
3. In RL, If the reward function is poorly designed, the agent may not learn the desired
behavior.
4. Reinforcement learning can be difficult to debug and interpret.
2.6.1 Types of Reinforcement:
1) Positive: Positive Reinforcement is defined as when an event, occurs due to a
particular behavior, increases the strength and the frequency of the behavior. In other
words, it has a positive effect on behavior.
 Advantages:
* Maximizes Performance
* Sustain Change for a long period of time.
* Too much Reinforcement can lead to an overload of states which can diminish
the results.
2) Negative: Negative Reinforcement is defined as strengthening of behavior because a
negative condition is stopped or avoided.
 Advantages:
* Increases Behavior.
* Provide defiance to a minimum standard of performance.
* It Only provides enough to meet up the minimum behavior.

2.6.2 Applications of Reinforcement Learning:

i) Robotics: Automating tasks in structured environments like manufacturing.
ii) Game Playing: Developing strategies in complex games like chess.
iii) Industrial Control: Real-time adjustments in operations like refinery controls.
iv) Personalized Training Systems: Customizing instruction based on individual
needs.
2.6.3 Illustration with Example: Navigating a Maze
The problem is as follows: We have an agent and a reward, with many hurdles in
between. The agent is supposed to find the best possible path to reach the reward. The
following problem explains the problem more easily.

44
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

The above image shows the robot, diamond, and fire. The goal of the robot is to get the
reward that is the diamond and avoid the hurdles that are fired. The robot learns by trying
all the possible paths and then choosing the path which gives him the reward with the
least hurdles. Each right step will give the robot a reward and each wrong step will subtract
the reward of the robot. The total reward will be calculated when it reaches the final
reward that is the diamond.

2.7 Comparison Table

Reinforcement
Criteria Supervised ML Unsupervised ML
ML
Trained using
Works on
Learns by using unlabeled data
Definition interacting with
labelled data without any
the environment
guidance.
No – predefined
Type of data Labelled data Unlabeled data
data
Type of Regression and Association and Exploitation or
problems classification Clustering Exploration
Extra
Supervision No supervision No supervision
supervision
Linear
Regression,
K – Means, Q – Learning,
Algorithms Logistic
C – Means, Apriori SARSA
Regression,
SVM, KNN etc.
Calculate Discover underlying Learn a series of
Aim
outcomes patterns action
Recommendation Self-Driving Cars,
Risk Evaluation,
Application System, Anomaly Gaming,
Forecast Sales
Detection Healthcare

45
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
 Department of CSE (AI & ML) II Year B.Tech-I Semester

Review Questions:
PART-A (TWO MARKS)

1. Mention the two fundamental logics in AI with any two key differences. Refer 1.3(C)

2. Write the basic elements of FOL syntax. Refer 1.1

3. State Generalized Modus Ponen’s rule with representation. Refer 1.2(C)

4. Define Unification and Lifting.Refer1.4,1.5

5. Write the Notation and Steps for Resolution. Refer 1.6

6. List the various Learning Methods in AI. Refer 2

7. Draw the schematic diagram for Decision tree and Explanation based learning. Refer 2.3,2.4

8. Define Observational learning and list the key elements. Refer 2.1

9. Give the diagrammatic representation for the Inductive Learning Algorithm with an example.
Refer 2.2

10. Write the key concepts of Reinforcement Learning. Refer 2.6

PART-B (TEN MARKS)

1. Define Logic in AI. Classify FOL and Propositional logic. Refer 1.3

2. Explain in detail about First order logic. Refer 1.2

3. What is inference in FOL? Discuss FOL Inference rules for quantifiers. Refer 1.2

4. Distinguish Forward and Backward chaining. Refer 1.7

5. Demonstrate briefly about various Learning Algorithms. Refer 2

6. Define Inductive Learning? Explain with an example. Refer 2.2

7. Elucidate the structure and working of Decision Trees with an example. Refer 2.3

8. Write the characteristics and working of Explanation-Based Learning. Refer 2.4

9. Discuss about various methods in Statistical Learning. Refer 2.5

46
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
 Department of CSE (AI & ML) II Year B.Tech-I Semester

10. Illustrate the working of Reinforcement Learning. Formulate Navigating a Maze problem. Refer
2.6,2.6.3

LECTURE PLAN
UNIT V
S.No. Topic No. of Proposed Actual Pertaining Taxonomy Mode of
Periods lecture lecture CO(s) Level Delivery

Period Period
1 Architecture 1 CO4 L2 MD1,
of expert MD5
systems
2 Roles of 1 CO4 L2 MD1,
expert MD5
systems
3 Knowledge 1 CO4 L2 MD1,
Acquisition MD5

4 Meta 1 CO4 L1 MD1,

knowledge MD5
Heuristics
5 Typical 1 CO4 L1 MD1,
expert MD5
systems–
MYCIN
6 DART 1 CO4 L1 MD1,
MD5
7 XCON: MD1,
Expert 1 CO4 L1 MD5
systems
shells.

47
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

UNIT - V
Expert Systems: Architecture of expert systems, Roles of expert systems – Knowledge
Acquisition Meta knowledge Heuristics. Typical expert systems – MYCIN, DART, XCON:
Expert systems shells.

1. Introduction
An expert system is an AI software that is designed to solve complex problems by using knowledge
from its knowledge base to provide decision-making ability like a human-expert. The expert
system is a part of AI, and the first ES was developed in the year 1970, which was the first
successful approach of artificial intelligence. One of the common examples of an ES is a
suggestion of spelling errors while typing in the Google search box.
1.1 Architecture of an Expert System
The process of building an expert system is often called knowledge engineering.

 Key Components of Expert System:

The components of an expert system typically include the knowledge base, inference
engine with explanation module, user interface, and learning module.
a) Knowledge Base:
The knowledge base stores/represents facts and rules. It consists of knowledge in
a particular domain as well as rules to solve a problem, procedures and intrinsic
data relevant to the domain.

48
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

1. Factual Knowledge − facts about a specific subject domain and it is accepted

by Knowledge Engineers (for example, Ramesh is an analyst).
2. Heuristic Knowledge (rules of thumb) − This knowledge is based on experience,
the capacity to guess, and evaluation. (for example, IF Ramesh is an analyst,
THEN he needs a workstation).
b) Inference Engine (Rules of Engine):
It applies inference rules to the knowledge base to derive conclusions or
recommendations. With the help of an inference engine, the system extracts the
knowledge from the knowledge base.
o Inference engines are divided into two types:
 Deterministic Inference Engine: In this form of an inference engine, the
conclusions are presumed to be correct. It is founded on facts and
regulations.
 Probabilistic Inference Engine: This sort of inference engine is based
on probability and contains uncertainty in findings.
o The following modes are used by the inference engine to generate
solutions:
i. Forward Chaining: This is a data-driven reasoning approach where
the system starts with available data and applies rules to deduce new
information until a conclusion is reached or no further inferences can
be made.
ii. Backward Chaining: This is a goal-driven reasoning approach where
the system starts with a goal or conclusion and works backward to
find evidence or data supporting that goal. It is useful for diagnostic
systems that need to identify the causes of observed symptoms.
c) User Interface
It is an interface that helps a non-expert user to interact with the expert system
to find a solution. It takes queries as an input in a readable format and passes it to
the inference engine. Example: Text-Based interface, Graphical User Interface
(GUI)
d) Knowledge Acquisition:
The function of this component is to allow the expert system to acquire more and
more knowledge from various sources and store it in the knowledge base.
e) Learning Module (Optional):

49
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Some expert systems incorporate a learning mechanism to improve performance.

Examples include supervised, unsupervised, reinforcement learning.
f) Explanation Module:
This module helps the expert system to give the user an explanation about how
the expert system reached a particular conclusion.
1.2 How Expert Systems Work?
Expert systems operate by following a structured approach:
1. Input Data: Users provide data or queries related to a specific problem or scenario.
2. Processing: The inference engine processes the input data using the rules in the
knowledge base to generate conclusions or recommendations.
3. Output: The system presents the results or solutions to the user through the user
interface.
4. Explanation: If applicable, the system explains how the conclusions were reached,
providing insights into the reasoning process.
1.3 Advantages of Expert Systems
1. Consistency in Decision-Making: Expert systems apply the same set of rules to
every problem.
2. 24/7 Availability: Expert systems can operate continuously, offering advice or
making decisions anytime.
3. Efficient Problem Solving: They can process large amounts of data quickly and
reach decisions faster than a human expert.
4. Cost-Effective: Once developed, an expert system can be a cost-effective solution for
organizations, as it reduces the need for constant human intervention.
5. Knowledge Preservation: Expert systems capture and store human expert
knowledge, which can be used over time.
1.4 Disadvantages of Expert Systems
1. Lack of Common Sense: Expert systems lack the ability to apply common sense or
judgment to unusual or novel situations.
2. Inability to Learn and Adapt: Traditional expert systems do not learn from new
experiences or evolve over time. Their knowledge base must be manually updated,
which can be labour-intensive and slow.
3. High Initial Development Costs: Building and implementing an expert system can
require significant upfront investment.
4. Dependence on Expertise: engineers are needed for the design and development of the

50
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

Expert system,
5. Limited Scope: They are not suitable for handling tasks that require human intuition,
creativity, or emotional intelligence.
1.5 Applications of Expert Systems
1. Medical Diagnosis: Expert systems assist doctors by analyzing symptoms and
medical history to suggest possible diagnoses or treatment options. For example,
MYCIN, an early expert system, helped identify bacterial infections and recommend
antibiotics.
2. Financial Services: In finance, expert systems are used for credit scoring, fraud
detection, and investment advice. They analyze financial data and patterns to make
informed decisions.
3. Technical Support: Expert systems can troubleshoot and provide solutions for
technical issues. They guide users through problem-solving steps based on pre-
defined rules and knowledge.
4. Manufacturing: In manufacturing, expert systems help optimize production
processes, perform quality control, and manage inventory by analyzing data and
making recommendations.

1.6 Roles of expert systems:

1. Knowledge Acquisition
Knowledge acquisition refers to the process of gathering, organizing, and structuring
knowledge from human experts or other sources to be used in an expert system. This
knowledge is often codified into rules or facts that the system uses to make decisions.

There are two main types of knowledge acquisition:

 Direct Knowledge Acquisition: Involves directly interviewing experts, reading
literature, or analysing databases to extract relevant information.
 Indirect Knowledge Acquisition: Uses automated tools, machine learning, or
other AI techniques to derive knowledge patterns from data or experiences.

51
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

The quality of the knowledge acquisition process directly affects the system's
performance and accuracy.
2. Meta-Knowledge
Meta-knowledge is "knowledge about knowledge." In expert systems, meta-knowledge
refers to the understanding of how the knowledge base is structured and how to use it
efficiently.

It involves knowing:
 How and when to apply specific rules.
 The strategies or methods to search and infer the best solution.
 The level of confidence in certain rules or pieces of knowledge.
Meta-knowledge helps an expert system manage uncertainty, optimize performance,
and prioritize actions. It provides context and reasoning guidelines for the system’s
knowledge.
3. Heuristics
Heuristics are rules of thumb or practical shortcuts that help in making decisions more
efficiently. They are not guaranteed to produce the optimal solution but often lead to
good enough solutions faster.

52
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

In expert systems, heuristics allow for:

 Efficient Problem Solving: By reducing the search space or focusing on the most
likely solutions.
 Decision-Making: Where complex reasoning may not be feasible or where time
constraints exist.
 Human-Like Reasoning: Mimicking how human experts use experience or
intuition to solve problems.
Expert systems use heuristics to simplify problem-solving processes, especially in
situations where complete information or resources may not be available.
1.6.1 Relationship Between Meta-Knowledge and Heuristics in Expert Systems
In expert systems, meta-knowledge and heuristics often work hand in hand. Meta-
knowledge can be used to decide when and how to apply certain heuristics to solve a
problem. For instance:
 Heuristics: Practical rules of thumb that simplify and speed up decision-making.
 Meta-Knowledge: Acts as a supervisory layer, guiding the inference engine in
using the knowledge base and heuristics effectively.
In this representation:
 The inference engine is at the core, interacting with both knowledge and
heuristics.
 Meta-knowledge monitors and optimizes the whole process, ensuring correct
usage of both rule-based knowledge and heuristic shortcuts.
 Examples of Meta-Knowledge and Heuristics in Expert Systems
 MYCIN (Medical Diagnosis): An early expert system used in medicine, MYCIN
employed heuristic rules for diagnosing bacterial infections and prescribing

53
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

treatments. Its meta-knowledge allowed it to assess the certainty of its

conclusions, providing users with confidence scores based on the available
evidence.
 XCON (Computer Configuration): XCON was used by Digital Equipment
Corporation to configure computer systems. It utilized heuristics for selecting
compatible hardware components based on user requirements and meta-knowledge to
manage conflicts between different rules.
1.7 Expert Systems/Shells:
The E.S shell simplifies the process of creating some knowledgebase. It is the shell that
processes the information entered by a user relates it to the concepts contained in the
knowledge base and provides an assessment or solution for a particular problem.

 Key Components of Expert System Shells:

1. Knowledge Base:
o Stores facts and rules related to the domain of expertise.
o The user inputs domain-specific knowledge in the form of if-then rules,
frames, or objects.
2. Inference Engine:
o The core component that applies logical rules to the knowledge base to
deduce new information or make decisions.
o Two main reasoning methods:
• Forward chaining: Starts with known facts and applies rules to
infer new facts until a goal is reached.
• Backward chaining: Starts with a goal and works backward to
see if there is evidence to support that goal.
3. User Interface:
o Provides interaction between the user and the system, typically through a
command-line or graphical interface.

54
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

o Allows users to input queries and receive explanations or advice.

4. Explanation Facility:
o Explains how the system reached a particular conclusion or decision,
helping users understand the reasoning process.
5. Knowledge Acquisition Tool:
o Assists in adding or updating the knowledge base, typically through a
guided interface for non-programmers.
 Advantages of Using Expert System Shells:
o Pre-built Components: Saves time by providing ready-made modules for
inference and knowledge management.
o No Need for Low-Level Programming: Users can focus on entering
domain knowledge without dealing with complex coding.
o Flexibility: Can be adapted to various domains, including medical diagnosis,
financial decision-making, and engineering.
 Limitations:
o Knowledge Engineering Required: Still requires expert knowledge to
define rules or facts.
o Performance Issues: Large knowledge bases can slow down processing in
complex systems.
o Lack of Creativity: Shells are limited to the rules and knowledge provided,
lacking the ability to "think" beyond them.
 Examples of Expert System Shells:
1. CLIPS (C Language Integrated Production System):
o An open-source shell designed for developing rule-based expert systems.
o Supports forward chaining and backward chaining.
2. EMYCIN (Empty MYCIN):
o A shell derived from the MYCIN expert system, used to build medical
and other rule-based systems.
3. Drools:
o A business rule management system (BRMS) that serves as an expert
system shell for creating decision automation solutions.
4. JESS (Java Expert System Shell):
o A Java-based shell used for building rule-based systems, widely used in
academia and industry.

55
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

1.8 Typical Expert Systems Shells:

Expert system shells like DART, MYCIN, and XCON represent early advancements in
the development of artificial intelligence, particularly in the field of expert systems.
1.8.1 DART (Decision Analysis and Resolution Tool)
 Purpose: DART is designed for decision support within expert systems, focusing
on multi-criteria decision analysis and resolution. It is more of a methodology or
tool than a standalone expert system, often integrated into broader decision
support systems or expert system shells.

 Functionality:
o Provides structured decision-making by analyzing different alternatives
based on weighted criteria.
o Supports decision-making processes in domains where multiple options
need to be evaluated against performance, risk, and other factors.
 Domain Applications:
DART can be applied in various fields like business management, engineering,
finance, and healthcare, where decisions must be made by comparing alternatives
and trade-offs.
 Key Features:
 Multi-criteria decision-making (e.g., cost vs. benefit analysis).
 Integration with inference engines and rule-based systems to automate decision-
making.
 Uncertainty handling and what-if analysis.
1.8.2 MYCIN
 Development: MYCIN was developed in the early 1970s at Stanford University
by Edward Shortliffe as a rule-based expert system focused on medical diagnosis
and treatment recommendations for infectious diseases, particularly bacterial
infections.

56
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Purpose: It was designed to assist physicians by providing recommendations on

which antibiotics to prescribe for patients, based on a set of symptoms and
laboratory results.
 Key Components of MYCIN Architecture:
 Knowledge Base: A set of rules encoding domain expertise.
 Inference Engine: Performs backward chaining to apply rules to the current
problem.
 User Interface: Interacts with the user through questions and answers.
 Explanation System: Provides the reasoning behind the system's conclusions.
 Certainty Handling: Manages uncertain data using certainty factors.

 Key Features:
 Rule-based inference engine (over 600 rules).
 Certainty factor-based reasoning to handle incomplete or uncertain data.
 Natural language processing for user interaction (though limited by the
technology at the time).
 Foundational for many later medical expert systems.
1.8.3 XCON (eXpert CONfigurer)
 Development: XCON (originally called R1) was developed by John McDermott
at Digital Equipment Corporation (DEC) in the late 1970s to solve a specific
business problem—configuring VAX computer systems.
 Purpose: XCON was designed to help DEC’s sales and manufacturing teams
configure the complex hardware components of their VAX computers, ensuring
that they were properly assembled and compatible.

57
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
Department of CSE (AI & ML) II Year B.Tech-I Semester

 Functionality:
o XCON is a rule-based system with hundreds of if-then rules, which automates
the process of configuring complex hardware systems by ensuring that all the
necessary components are included and correctly interconnected.
o It applies forward chaining—starting from the facts provided by the user
(e.g., specific hardware parts required) and then working forward through rules
to arrive at a configuration solution.
 Domain Applications: Primarily used in the manufacturing and sales domains
for configuring complex systems, especially for computer hardware and network
design.
 Key Features:
 Rule-based system designed for complex configuration tasks.
 Forward chaining inference engine to apply knowledge rules to specific facts.
 Used a large set of rules (eventually over 10,000) to cover a wide range of
hardware configurations.
 Ensured that systems were error-free and complete, saving DEC time and
reducing human errors in configuration.
1.8.4 Comparison of DART, MYCIN, and XCON:

Feature DART MYCIN XCON (R1)

Rule-Based
Decision Analysis and Rule-Based Medical
Type Configuration Expert
Resolution Tool Expert System
System
Medicine
Primary General Decision Manufacturing and Sales
(Infectious
Domain Support (Computer Systems)
Diseases)

Inference Multi-Criteria Decision Backward Chaining Forward Chaining (Rule-

Method Making (Rule-Based) Based)

Handling Probabilistic and multi- Limited (Focuses on

Certainty Factors
Uncertainty criteria Models Configuration)

Diagnosis and
Primary Use VAX Computer
Decision Optimization Treatment of
Case Configuration
Infections

Enhanced decision-
Pioneering medical Successful commercial
Impact making in various
AI AI system
domains

58
Compiled By: P. Rajyalakshmi S.V.P.C.E.T
 Department of CSE (AI & ML) II Year B.Tech-I Semester

Review Questions:
Part-A (Two Marks)
Define Expert system.

Write the advantages and disadvantages of expert systems.

List the components of expert systems.

Define inference engine.

What is knowledge base?

What is meant by knowledge acquisition?

Compare two different expert system shells (e.g., CLIPS and Jess)
Write the functionality of XCON.
Give the architecture of MYCIN in medical diagnosis.
Draw the schematic diagram for expert system system shell.

Part-B (Ten Marks)

With a neat sketch explain the architecture of expert system.

Explain the role of the knowledge base and inference engine in the architecture of an expert system.

Discuss the role of knowledge acquisition in Expert System.

What is meta-knowledge in expert systems? Provide the functionality and examples.

How do heuristics improve the performance of expert systems? Provide examples from real-world
applications.

How do meta-knowledge and heuristics work together in real-world scenarios.

Examine the architecture and functionality of the MYCIN expert system in medical diagnosis.

Discuss the role of the DART in expert system.

Compare and contrast MYCIN, DART, and XCON expert systems

Explain the key components of an expert system shell and discuss the benefits and limitations.

59
Compiled By: P. Rajyalakshmi S.V.P.C.E.T