Artificial Intelligence and Expert Systems E-Book

Mercury Learning and Information

• Herausgeber: Packt Publishing
• Kategorie: Fachliteratur
• Sprache: Englisch

Seitenzahl: 472

Veröffentlichungsjahr: 2024

Beschreibung

This book covers current AI applications and techniques for solving problems and accomplishing tasks. It introduces branches of AI such as formal logic, reasoning, knowledge engineering, expert systems, neural networks, and fuzzy logic. It emphasizes expert systems, with sections on state space search, knowledge engineering, neural networks, fuzzy logic, and Prolog.
It begins with an introduction to AI and its applications, setting the stage for foundational concepts. Readers are guided through state space search and heuristic search strategies, crucial for problem-solving in AI. The focus shifts to expert systems, covering their development life cycle, knowledge acquisition, and representation, providing a deep dive into emulating human decision-making.
Later chapters cover neural networks and the learning process, essential for creating adaptive systems. Sections on fuzzy logic and fuzzy systems introduce methods for handling uncertainty in AI. Final chapters on programming in logic and advanced Prolog offer practical techniques for AI solutions. This approach equips readers with the skills to apply AI in various domains, enhancing their problem-solving abilities and understanding of intelligent systems.

Leseprobe

ARTIFICIAL INTELLIGENCE AND EXPERT SYSTEMS

Itisha Gupta&Garima Nagpal

MERCURY LEARNING AND INFORMATION

Dulles, VirginiaBoston, MassachusettsNew Delhi

Copyright © 2020 by MERCURY LEARNING AND INFORMATION.All rights reserved.

Original title and copyright: Artificial Intelligence and Expert System.Copyright ©2018 by Laxmi Publications Pvt. Ltd. All rights reserved.

This publication, portions of it, or any accompanying software may not be reproduced in any way, stored in a retrieval system of any type, or transmitted by any means, media, electronic display or mechanical display, including, but not limited to, photocopy, recording, Internet postings, or scanning, without prior permission in writing from the publisher.

Publisher: David PallaiMERCURY LEARNING AND INFORMATION22841 Quicksilver DriveDulles, VA [email protected]

I. Gupta & G. Nagpal. Artificial Intelligence and Expert Systems.ISBN: 978-1-68392-507-1

The publisher recognizes and respects all marks used by companies, manufacturers, and developers as a means to distinguish their products. All brand names and product names mentioned in this book are trademarks or service marks of their respective companies. Any omission or misuse (of any kind) of service marks or trademarks, etc. is not an attempt to infringe on the property of others.

Library of Congress Control Number: 2020935416

202122321                            Printed on acid-free paper in the United States of America.

Our titles are available for adoption, license, or bulk purchase by institutions, corporations, etc. For additional information, please contact the Customer Service Dept. at 800-232-0223(toll free).

All of our titles are available in digital format at academiccourseware.com and other digital vendors. The sole obligation of MERCURY LEARNING AND INFORMATION to the purchaser is to replace the book, based on defective materials or faulty workmanship, but not based on the operation or functionality of the product.

CONTENTS

Preface

Chapter 1:Introduction to Artificial Intelligence

1.1.The Turing Test

1.2.Intelligent Agents

1.2.1.Software Agents

1.2.2.Physical Agents

1.3.Approaches in Artificial Intelligence

1.3.1.Acting Humanly: The Turing Test Approach

1.3.2.Thinking Humanly: The Cognitive Modelling Approach

1.3.3.Thinking Rationally: The Laws of Thought Approach

1.3.4.Acting Rationally: The Rational Agent Approach

1.4.Definitions of Artificial Intelligence

1.4.1.Intelligent Behavior

1.4.2.Interpretations of Artificial Intelligence

1.5.AI Problems

1.5.1.Tasks Under Artificial Intelligence

1.5.2.Tasks Domains of Artificial Intelligence

1.6.Features of AI Programs

1.7.Importance of AI

1.8.What Can Artificial Intelligence Systems Do?

1.9.What Can Artificial Intelligence Systems Not Do Yet?

1.10.Advantages of AI

1.11.Disadvantages of Artificial Intelligence

Exercises

Chapter 2:Applications of Artificial Intelligence

2.1.Finance

2.2.Hospitals and Medicine

2.3.Robotics

2.4.Expert Systems

2.5.Diagnosis

2.6.Pattern Recognition

2.7.Natural Language Processing

2.8.Game Playing

2.9.Image Processing

2.10.Data Mining

2.11.Big Data Mining

Exercises

Chapter 3:Introduction to the State Space Search

3.1.State Space Search

3.1.1.The Search Problem

3.2.Search Techniques

3.2.1.Basic Search Algorithm

3.3.Types of Searching Techniques

3.3.1.Uninformed Search (Blind Search)

3.3.2.Avoiding Repeated States

Exercises

Chapter 4:Heuristic Search Strategies

4.1.Types of Heuristic Search Techniques

4.1.1.Generate and Test

4.1.2.Best First Search

4.1.3.Hill Climbing Search

4.1.4.Simulated Annealing Search

4.1.5.A∗ Algorithm

4.1.6.AND-OR Graphs

4.2.Properties of the Heuristic Search Algorithm

4.3.Adversary Search

4.3.1.The MINIMAX Algorithm

Exercises

Chapter 5:Expert Systems

5.1.Definitions of Expert Systems

5.2.Features of Good Expert Systems

5.3.Architecture and Components of Expert Systems

5.3.1.User Interface

5.3.2.Knowledge Base

5.3.3.Working Storage (Database)

5.3.4.Inference Engine

5.3.5.Explanation Facility

5.3.6.Knowledge Acquisition Facility

5.3.7.External Interface

5.4.Roles of the Individuals Who Interact with the System

5.4.1.Domain Expert

5.4.2.Knowledge Engineer

5.4.3.Programmer

5.4.4.Project Manager

5.4.5.User

5.5.Advantages of Expert Systems

5.6.Disadvantages of Expert Systems

Exercises

Chapter 6:The Expert System Development Life Cycle

6.1.Stages in the Expert System Development Life Cycle

6.1.1.Problem Selection

6.1.2.Conceptualization

6.1.3.Formalization

6.1.4.Prototype Construction

6.1.5.Implementation

6.1.6.Evaluation

6.2.Sources of Error in Expert System Development

6.2.1.Knowledge Errors

6.2.2.Syntax Errors

6.2.3.Semantic Errors

6.2.4.Inference Engine Errors

6.2.5.Inference Chain Errors

Exercises

Chapter 7:Knowledge Acquisition

7.1.Knowledge Basics

7.2.Knowledge Engineering

7.2.1.Knowledge Acquisition

7.2.2.Knowledge Engineer

7.2.3.Difficulties in Knowledge Acquisition

7.3.Knowledge Acquisition Techniques

7.3.1.Natural Techniques

7.3.2.Contrived Techniques

7.3.3.Modelling Techniques

Exercises

Chapter 8:Knowledge Representation

8.1.Definitions of Knowledge Representation

8.2.Characteristics of Good Knowledge Representation

8.3.Basics of Knowledge Representation

8.4.Properties of the Symbolic Representation of Knowledge

8.5.Properties for the Good Knowledge Representation Systems

8.6.Categories of Knowledge Representation Schemes

8.7.Types of Knowledge Representational Schemes

8.7.1.Formal Logic

8.7.2.Semantic Net

8.7.3.Frames

8.7.4.Scripts

8.7.5.Conceptual Dependency (CD)

Exercises

Chapter 9:Neural Networks

9.1.Neural Networks vs. Conventional Computers

9.2.Neural Networks

9.2.1.Neurons

9.2.2.Types of Neural Networks

9.2.3.Historical Background

9.3.Biological Neural Networks

9.3.1.Biological Neurons

9.4.Artificial Neural Networks

9.5.Differences Between Biological and Artificial Neural Networks

9.6.Architecture of a Neural Network

9.6.1.Single Layer Feed-Forward Networks

9.6.2.Multilayer Feed-Forward Network

9.6.3.Recurrent Networks

9.6.4.Feedback Networks

9.6.5.Network Layers

Exercises

Chapter 10:The Learning Process

10.1.Types of Learning in a Neural Network

10.1.1.Supervised Learning

10.1.2.Unsupervised Learning

10.1.3.Reinforcement Learning

10.2.Perceptron

10.2.1.The Representational Power of a Perceptron

10.3.Backpropagation Networks

10.4.Advantages of Neural Networks

10.5.Limitations of Neural Networks

10.6.Applications of Neural Networks

Exercises

Chapter 11:Fuzzy Logic

11.1.Introduction to Fuzzy Logic

11.1.1.Definition of Fuzzy Logic

11.1.2.Features of Fuzzy Logic

11.1.3.Advantages of Fuzzy Logic

11.1.4.Disadvantages of Fuzzy Logic

11.2.Crisp Set (Classical set)

11.3.Fuzzy Set

11.3.1.Linguistic Variables in a Fuzzy Set

11.4.Membership Function of Crisp Logic

11.5.Membership Function of the Fuzzy Set

11.6.Fuzzy Set Operations

11.6.1.Union

11.6.2.Intersection

11.6.3.Complement

11.6.4.Equality of Two Fuzzy Sets

11.6.5.Containment

11.6.6.Normal Fuzzy Set

11.6.7.Support of a Fuzzy Set

11.6.8.α-Cut or α-Level Set

11.6.9.Disjunctive Sum (Exclusive OR)

11.6.10.Disjoint Sum

11.6.11.Difference

11.6.12.The Bounded Difference

11.7.Properties of A Fuzzy Set

11.8.Differences Between a Fuzzy Set and A Crisp Set

11.9.Differences Between Boolean Logic and Fuzzy Logic

Exercises

Chapter 12:Fuzzy Systems

12.1.Fuzzy Rule

12.1.1.Fuzzy Rules as Relations

12.1.2.Interpretation of Fuzzy Rules

12.2.Fuzzy Reasoning

Exercises

Chapter 13:Fuzzy Expert Systems

13.1.The Need for Fuzzy Expert Systems

13.2.Operations on a Fuzzy Expert System

13.2.1.Fuzzification (Fuzzy Input)

13.2.2.Fuzzy Operator

13.2.3.Fuzzy Inferencing (Implication)

13.2.4.Aggregate All Output

13.2.5.Defuzzification

13.3.Fuzzy Inference Systems

13.3.1.Mamdani Fuzzy Inference Method

13.3.2.Sugeno Inference Method (TSK Fuzzy Model of Takagi, Sugeno, and Kang)

13.3.3.Choosing the Inference Method

13.4.The Fuzzy Inference Process in a Fuzzy Expert System

13.4.1.Monotonic Inference

13.4.2.Non-Monotonic Inference

13.4.3.Downward Monotonic Inference

13.5.Types of Fuzzy Expert Systems

13.5.1.Fuzzy Control

13.5.2.Fuzzy Reasoning

13.6.Fuzzy Controller

13.6.1.Components of a Fuzzy Controller

13.6.2.Application Areas of Fuzzy Controller

Exercises

Chapter 14:Logic Programming

14.1.Introduction

14.2.Difference Between C/C++ and Prolog

14.3.How Does Prolog Work?

14.4.A Little History

14.5.Converting English to Prolog

14.6.Goals

14.6.1.How Prolog Satisfies Goals

14.7.Queries

14.8.Clauses

14.8.1.Facts

14.8.2.Rules

14.9.Notation in Prolog for Building Blocks

14.9.1.Atoms

14.9.2.Variables

14.9.3.Data Types and Structures

14.10.Arithmetic Operations

14.11.Strings

Exercises

Chapter 15:Advanced Prolog

15.1.Input and Output Predicates

15.1.1.Terms and Character I/O

15.1.2.File I/O

15.2.Backtracking

15.2.1.Problems with Backtracking

15.3.Cut

15.4.Fail

15.4.1.Cut and Fail Combination

15.5.Recursion

15.6.Prolog Data Structure

15.6.1.Terms

15.6.2.Unification

15.7.Dynamic Database

15.8.Programs in Prolog

15.9.Problems with Prolog

Exercises

Index

PREFACE

Artificial Intelligence (AI) is a branch of computer and information science. The goal of AI is to create a machine that behaves like an ordinary human with an improved machine behavior in tackling complex tasks and to accomplish those tasks in such a way that they would be considered to display “intelligence.”

Artificial Intelligence and Expert Systems is a book about the science of artificial intelligence. It is designed to help readers in learning about some of the current applications and techniques of AI as an aid to solving problems and accomplishing tasks. The book provides a general introduction to problems and techniques of AI. We have tried to explore the various branches of AI, which encompass formal logic, reasoning, knowledge engineering, expert system neural networks, fuzzy logic, etc. Thus, this book has been structured into parts that benefit the reader in choosing from a variety of paths to the chapters. This book is divided into five parts: problems and state space, knowledge engineering, neural networks, fuzzy logic, and Prolog.

Part I provides introductory concepts on various problems that AI seeks to solve. It introduces a coherent framework in which to understand AI. It also includes a detailed explanation of various state space search algorithms such as best first search, hill climbing, A* algorithms, and uniform search techniques.

Reasoning is one of the important fields of AI that requires a great deal of knowledge about the world in order to solve complex problems and simulates the decision-making ability of man. Part II explores an important application of AI, i.e. the field of expert systems which was among the first truly successful forms of AI software, designed to solve complex problems by reasoning, like an expert solving complex tasks. This part also introduces the various methods of knowledge acquisition from human domain experts and explores various knowledge representational schemes like predicate logic, propositional logic, frames, scripts and semantic networks.

Part III describes another important branch of AI that is neural networks which are simplified models of the biological neuron system. Neural networks are a parallel distributed processing system that is made by highly interconnected computing elements, used to learn and thereby acquire knowledge. This part provides a detailed explanation of the various forms of artificial neural networks, e.g., single layer feed, forward neural networks, multilayer networks, feed backward networks and various learning methods including supervised, unsupervised, and reinforcement.

Part IV provides basic concepts on fuzzy logic introduced in 1930s by Jan Lukasiewicz. Fuzzy logic is a problem-solving, control system methodology that is used in systems ranging from simple, small, embedded micro-controllers to large, networked, multi-channel PC or workstation-based data acquisition and control systems. Fuzzy logic provides a simple way to arrive at a definite conclusion based upon vague, ambiguous, imprecise, noisy, or missing input information. It also explores the reasoning process in fuzzy logic to derive conclusions from known facts and rules. Further, it provides introductory concepts on a fuzzy expert system to deal with uncertainty and ambiguities that are difficult to deal with in conventional expert systems.

Part V describes Prolog which is a programming language of AI with an ultimate goal of developing code for solving AI problems. Prolog is a declarative (descriptive) language, non-procedural in nature. The programs are written in a way that not only defines how the computational process is to be carried out, but also consists of several declarations representing significant facts and rules. The solution to be mined is also expressed as a question to be answered and a goal to be achieved.

CHAPTER 1

INTRODUCTION TO ARTIFICIAL INTELLIGENCE

We all know that computers are suitable for performing mechanical computations using fixed programmed rules that allow machines to perform simple monotonous tasks efficiently and reliably. Human beings get bored very quickly with monotonous tasks. A computer cannot reason and lacks common sense, and it is difficult for a computer to understand new situations and adapt itself. However, human beings can adapt themselves to new situations since they have the ability to reason. Human beings see through their eyes, and their brains interpret this input to extract the types of objects in the scene. A human being hears a set of voice signals through their ears, and the brain interprets it as a meaningful sentence. Thus, the goal of Artificial Intelligence (AI) is to create a machine that behaves like an ordinary human being and that is an improvement over current machine behavior for tackling complex tasks.

Much of AI research has allowed us to understand our intelligent behavior. Humans have an interesting approach to problem-solving that is based on abstract thought, high-level deliberative reasoning, and pattern recognition. AI can help us understand this process by recreating it, enabling us to enhance our abilities. AI currently includes a huge variety of subfields, from general-purpose areas such as perception and logical reasoning, to specific tasks, such as playing chess, proving mathematical theorems, writing poetry, and diagnosing diseases. Often, scientists in other fields move gradually into artificial intelligence, where they find the tools and vocabulary to systematize and automate the intellectual tasks on which they have been working all their lives.

Although artificial intelligence as an independent field of study is relatively new, it has some roots in the past. We can say that it started 2,400 years ago when the Greek philosopher Aristotle invented the concept of logical reasoning. The effort to finalize the language of logic continued with Leibniz and Newton. George Boole developed Boolean algebra in the nineteenth century, which laid the foundation of computer circuits. However, the main idea of a thinking machine came from Alan Turing, who proposed the Turing test. In 1950, Alan Turing proposed the Turing test, which provides a definition of intelligence in a machine. The term “artificial intelligence” was first coined by John McCarthy in 1956.

1.1The Turing Test

The English mathematician Alan M. Turing devised a test to determine whether a computer can be said to think like a human. The test was named after Turing, who founded artificial intelligence during the 1940s and 1950s. The original version of the test asked the question “Can machines think?” According to this test, a computer is deemed to have artificial intelligence if it can mimic human responses under specific conditions. In Turing’s test, if the human conducting the test is unable to consistently determine whether an answer has been given by a computer or by another human being, then the computer is considered to have “passed” the test. In the basic Turing test, there are three terminals. Two of the terminals are operated by humans, and the third terminal is operated by a computer. Each terminal is physically separated from the other two. One human is designated as the questioner (interrogator). The other human and the computer are designated the respondents. The questioner interrogates both the human respondent and the computer according to a specified format, within a certain subject area and context, and for a pre-set length of time (such as 10 minutes). The test simply compares the intelligent behavior of a human being with that of a computer. An interrogator asks a set of questions that are forwarded to both the computer and the human. The interrogator receives two sets of responses, but does not know which set comes from the human and which set from the computer. After a careful examination of the two sets, if the interrogator cannot definitely tell which set has come from the computer and which from the human, the computer has passed the Turing test for intelligent behavior. However, the test is not as straight-forward as it seems because humans are superior to computers in creativity, common sense, and reasoning. If the test uses any question that is related to these concepts, then the human is sure beat the computer. Computers are more accurate and faster at performing computations than humans. The Turing test is a test that a machine should pass in order to be called intelligent.

FIGURE 1.1 An example of the Turing test, in which the Interrogator must determine which respondent is the computer

There are some criticisms of the Turing test:

•A machine could pass the Turing test, but it is a different matter to know what the level of proficiency the machine actually has.

•Searle proposed an argument called the “Chinese room argument” to bring attention to a major flaw in the Turing test. According to this argument, Searle did not know Chinese and was locked in room with set of Chinese letters. He was given some writing in Chinese with instructions in English that correlated to the first and second set of symbols. He was also given a set of questions for answering (which was supplemented by instructions in English). He claimed that he could manipulate the Chinese symbols in a formal way and provide a satisfactory answer to people outside the room that would create the illusion that he knew Chinese. Searle argued that a machine that passes the Turing test and is assumed to be intelligent actually behaves in the same fashion (it manipulates formal symbols with a lack of understanding).

•The Turing test has been criticized because the nature of the questioning must be limited in order for a computer to exhibit human-like intelligence. For example, a computer might score high when the questioner formulates the queries so they have “Yes” or “No” answers and pertain to a narrow field of knowledge, such as mathematical number theory. If the responses to the questions are of a broad-based, conversational nature, however, a computer would not be expected to perform like a human being. This is especially true if the subject is emotionally charged or socially sensitive.

•In some specialized instances, a computer may perform so much better and faster than a human that the questioner can easily tell which is which. Google and Yahoo are examples of computer applications that outperform a human in a Turing test based on information searches.

These arguments highlight the deficiencies of the Turing test and raise the question “What is intelligence?”

Intelligence: This is the ability to reason, develop new thoughts, perceive, and learn. Psychologists have proposed various definitions, but there is no consensus on any particular definition.

The term “thought” can be defined as a mechanism which

a)stimulates

•action

•information generation

•knowledge generation

b)is triggered by

•external stimulus

•internal stimulus

c)acts through

•present environment

•past memory

d)is stored as

•the charged/discharged state of neurons

•electromagnetic thought waves

1.2Intelligent Agents

An intelligent agent is a system that perceives its environment, learns from it, and interacts with it intelligently. Intelligent agents can be divided into two broad categories: software agents and physical agents.

1.2.1Software Agents

A software agent is a set of programs that is designed to do particular tasks. For example, a software agent can check the contents of received e-mails and classify them into different categories (junk, less important, important, very important, and so on). Another example of a software agent is a search engine used to search the World Wide Web and find sites that can provide information about a requested subject.

1.2.2Physical Agents

A physical agent (robot) is a programmable system that can be used to perform a variety of tasks, e.g., simple robots can be used in manufacturing industries for performing various routine jobs such as assembling, welding, or painting. Some organizations use mobile robots for performing routine delivery jobs, such as distributing mail or correspondence to different rooms. Mobile robots are used underwater to prospect for oil.

AI is the branch of computer science which aims to make computers behave like human beings. The term was coined in 1956 by John McCarthy at the Massachusetts Institute of Technology. The various areas of artificial intelligence include

•game playing:programming computers to play games, such as chess and checkers

•expert systems: programming computers to make decisions in real-life situations (for example, some expert systems help doctors diagnose diseases based on symptoms)

•natural language: programming computers to understand natural human languages

•neural networks: systems that simulate intelligence by attempting to reproduce the types of physical connections that occur in animal brains

•robotics: programming computers to see and hear and react to other sensory stimuli.

There are many different approaches to artificial intelligence, none of which are completely right or wrong. Through the years, new techniques have emerged based on the state of mind of the researchers, funding opportunities, and the available computer hardware.

Over the past five decades, AI research has mostly focused on solving specific problems. Many solutions have been proposed, and there have been improvements in the efficiency and reliability of these solutions. AI is divided into many fields, ranging from pattern recognition to artificial life. AI is a broad discipline that simulates human skills such as automatic programming, case-based reasoning, neural networks, decision-making, expert systems, natural language processing, pattern recognition, and speech recognition. AI technologies bring more complex data analysis features to existing applications.

Currently, no computers exhibit full artificial intelligence (a simulation of human behavior). The greatest advances have occurred in the field of game playing. The best computer chess programs are now capable of beating humans. In May, 1997, an IBM super-computer called Deep Blue defeated world chess champion Gary Kasparov in a chess match. Computers are now widely used in assembly plants, but they are capable only of very limited tasks. Robots have great difficulty identifying objects based on appearance or feel, and they still move and handle objects clumsily.

There are several programming languages that are known as AI languages because they are used almost exclusively for AI applications. The two most common are LISP and Prolog.

1.0 Artificial Intelligence Characteristics

Systems that act like humans

Systems that act rationally

1.3Approaches in Artificial Intelligence

1.3.1Acting Humanly: The Turing Test Approach

AI is a system that “thinks” like humans and can be explained using the Turing test. The Turing Test was designed to provide a satisfactory operational definition of intelligence. Turing defined intelligent behavior as the ability to achieve human-level performance in all cognitive tasks, sufficient to fool an interrogator. The test he originally proposed is that the computer should be interrogated by a human via a teletype, and the computer passes the test if the interrogator cannot tell if there is a computer or a human at the other end. The computer would need to possess the following capabilities to pass the Turing test:

•natural language processing to enable it to communicate successfully in English (or some other human language)

•knowledge representation to store information provided before or during the interrogation

•automated reasoning to use the stored information to answer questions and draw new conclusions

•machine learning to adapt to new circumstances and to detect and extrapolate patterns

Turing’s test deliberately avoided direct physical interaction between the interrogator and the computer because the physical simulation of a person is unnecessary for intelligence. However, the 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

•robotics to move them about

Within AI, there has not been a big effort to try to pass the Turing test. The issue of acting like a human comes up primarily when AI programs have to interact with people, as when an expert system explains how it came to its diagnosis or a natural language processing system has a dialogue with a user. These programs must behave according to certain normal conventions of human interactions in order to make themselves understood. The underlying representation and reasoning in such a system may or may not be based on a human model.

1.3.2Thinking Humanly: The Cognitive Modelling Approach

If we are going to say that a given program thinks like a human, we must have some way of determining how humans think. We need to get inside the actual workings of human minds. There are two ways to do this: through introspection—trying to catch our own thoughts as they go by—or through psychological experiments. Once we have a sufficiently precise theory of the mind, it becomes possible to express the theory as a computer program. If the program’s input/output and timing behavior matches human behavior, that is evidence that some of the program’s mechanisms may also be operating in humans. The interdisciplinary field of cognitive science brings together computer models from AI and experimental techniques from psychology to try to construct precise and testable theories of the workings of the human mind. Real cognitive science, however, is necessarily based on the experimental investigation of actual humans or animals, and we assume that the reader only has access to a computer for experimentation. We will simply note that AI and cognitive science continue to enrich each other, especially in the areas of vision, natural language, and learning.

1.3.3Thinking Rationally: The Laws of Thought Approach

The Greek philosopher Aristotle was one of the first to attempt to codify “right thinking,” that is, unquestionable reasoning processes. His famous syllogisms provided patterns for argument structures that always gave correct conclusions given correct premises. For example, “Socrates is a man; all men are mortal; therefore, Socrates is mortal.” These laws of thought were supposed to govern the operation of the mind and initiated the field of logic.

The development of formal logic in the late nineteenth and early twentieth centuries, which we describe in more detail in the next chapters, provided a precise notation for statements about all kinds of things in the world and the relationships between them. By 1965, programs existed that could, given enough time and memory, take a description of a problem in logical notation and find the solution to the problem, if one existed. (If there is no solution, the program might never stop looking for it.) There are two main obstacles to this approach. First, it is not easy to take informal knowledge and state it in the formal terms required by logical notation, particularly when the knowledge is less than 100% certain. Second, there is a big difference between being able to solve a problem in principle and doing so in practice.

1.3.4Acting Rationally: The Rational Agent Approach

Acting rationally means acting so as to achieve one’s goals given one’s beliefs. An agent is just something that perceives and acts. (This may be an unusual use of the word, but you will get used to it.) In this approach, AI is viewed as the study and construction of rational agents.

In the “laws of thought” approach to AI, the whole emphasis was on correct inferences. Making correct inferences is sometimes part of being a rational agent, because one way to act rationally is to reason logically to the conclusion that a given action will achieve one’s goals, and then to act on that conclusion. On the other hand, correct inference is not all of rationality; because there are often situations where there is no provably correct thing to do, yet something must still be done. There are also ways of acting rationally that cannot be reasonably said to involve inference. For example, pulling one’s hand off of a hot stove is a reflex action that is more successful than a slower action taken after careful deliberation.

All the “cognitive skills” needed for the Turing test are there to allow for rational actions. Thus, we need the ability to represent knowledge and reason with it because this enables us to reach good decisions in a wide variety of situations. We need to be able to generate comprehensible sentences in natural language because saying those sentences helps us get by in a complex society. We need learning not just for erudition, but because having a better idea of how the world works enables us to generate more effective strategies for dealing with it. We need visual perception not just because seeing is fun, but in order to get a better idea of what an action might achieve—for example, being able to see a tasty morsel helps one to move toward it.

The study of AI as the design of a rational agent therefore has two advantages. First, it is more general than the “laws of thought” approach because correct inference is only a useful mechanism for achieving rationality, and not a necessary one. Second, it is more amenable to scientific development than approaches based on human behavior or human thought because the standard of rationality is clearly defined and completely general. Human behavior, on the other hand, is well-adapted for one specific environment and is the product, in part, of a complicated and largely unknown evolutionary process that still may be far from achieving perfection.

1.4Definitions of Artificial Intelligence

A number of definitions have been proposed for AI:

•AI is a technology and a branch of computer science that studies and develops intelligent machines and software.

•Software technologies that make a computer or robot perform equal or better than normal human computational ability in accuracy, capacity, and speed.

•Artificial intelligence is a branch of science which deals with helping machines find solutions to complex problems in a human-like fashion. This generally involves borrowing characteristics from human intelligence and applying them as algorithms in a computer-friendly way.

•Artificial intelligence is the study of programmed systems that can simulate, to some extent, human activities such as perceiving, thinking, learning, and acting.

•“The art of creating machines that perform functions that require intelligence when performed by people.” (Kurzweil, 1990)

•“The study of how to make computers do things at which, at the moment, people are better.” (Rich and Knight, 1991)

•“A field of study that seeks to explain and emulate intelligent behavior in terms of computational processes.” (Schalkoff, 1990)

•“The branch of computer science that is concerned with the automation of intelligent behavior.” (Luger and Stubblefield, 1993).

Artificial intelligence is concerned with the design of intelligence in an artificial device. The term was coined by McCarthy in 1956.

There are two ideas in the definition.

1.Intelligence

2.Artificial device

What is Intelligence?

•Intelligence is the computational part of the ability to achieve goals in the world. Varying kinds and degrees of intelligence occur in people, many animals, and some machines.

•The capacity to learn and solve problems

In particular,

•the ability to solve novel problems

•the ability to act rationally

•the ability to act like humans

♦A system with intelligence is expected to behave as intelligently as a human

♦A system with intelligence is expected to behave in the best possible manner

What is involved in intelligence?

•The ability to interact with the real world

♦to perceive, understand, and act

♦e.g., speech recognition and understanding and synthesis

♦e.g., image understanding

♦e.g., the ability to take actions, to have an effect

•Reasoning and Planning

♦modeling the external world, given input

♦solving new problems, planning, and making decisions

♦the ability to deal with unexpected problems and uncertainties

•Learning and Adaptation

♦We are continuously learning and adapting.

♦Our internal models are always being “updated.”

♦One example is a baby learning to categorize and recognize animals.

1.4.1Intelligent Behavior

Tasks and applications that constitute intelligent behavior are

•perception involving image recognition and computer vision

•reasoning

•learning

•understanding language involving natural language processing and speech processing

•solving problems

•robotics

1.4.2Interpretations of Artificial Intelligence

Different interpretations have been used by researchers for defining the scope and view of AI.

a)One view is that artificial intelligence is about designing systems that are as intelligent as humans. This view means that we should try to understand human thought and build machines that simulate the human thought process. This view is the cognitive science approach to AI.

b)The second approach is best defined by the concept of the Turing test. The Turing test is a kind of imitation game, in which a human being and a computer are interrogated under conditions where the interrogator does not know which is machine and which is human. The communications are carried out entirely via text messages. Turing argued that if the interrogator could not distinguish them by questioning, then it would be unreasonable not to call the computer intelligent. Turing’s imitation game is the Turing test.

c)The third view of AI is that it is the study of rational agents. This view deals with building machines that act rationally. The focus is on how the system acts and performs, and not so much on the reasoning process. A rational agent is one that acts rationally, that is, is in the best possible manner.

1.5AI Problems

While studying the typical range of tasks that we might expect an intelligent entity to perform, we need to consider both common-place tasks as well as expert tasks.

a)A lot of work in AI is focused on formal tasks such as game playing and theorem proving. Samuel wrote a checker playing program that not only plays the game with opponents, but also uses its experience to improve its later performance. Such types of tasks require intelligence, so people who do these tasks well are considered intelligent. It may seem like a computer could perform such tasks well by exploring a large number of solution paths fast and select best one. But this assumption is false, since no computer is fast enough to overcome the combinatorial explosion generated by most problems.

b)Another focus in AI is solving everyday tasks that require common sense reasoning. This includes reasoning about physical objects and their relationship to each other, as well as reasoning about actions and consequences.

c)As AI research has progressed, techniques have been developed for handling a large amount of knowledge. Progress was made in handling more complex tasks such as perception, natural language understanding, and diagnosis problems.

d)Animal have less intelligence than humans, but have more sophisticated visual perception. Perceptual tasks are difficult because they involve analog signals (noisy signals).

e)Natural language understanding is a problem. In addition to these mundane tasks, people may perform one or more specialized tasks that require expertise, such as engineering design tasks, medical diagnosis, and scientific discovery tasks.

1.5.1Tasks Under Artificial Intelligence

There are tasks done routinely by humans and animals. Examples of common-place tasks include

•recognizing people and objects

•communicating (through natural language)

•navigating around obstacles on the streets

Examples of expert tasks include

•medical diagnosis

•mathematical problem solving

•playing games like chess

Expert tasks cannot be done by all people; they can only be performed by skilled specialists. Clearly tasks of the first type are easy for humans to perform, and almost all are able to master them. The second range of tasks requires skill development and/or intelligence. Only some specialists can perform them well. The achievements of computer systems include performing sophisticated tasks like making a medical diagnosis, performing symbolic integration, proving theorems, and playing chess.

However, it has proven to be very difficult to make computer systems perform many routine tasks that all humans and a lot of animals can do. Examples of such tasks include navigating our way without running into things, catching prey, and avoiding predators. Humans and animals are also capable of interpreting complex sensory information. We are able to recognize objects and people from the visual image that we receive. We are also able to perform complex social functions.

1.5.2Tasks Domains of Artificial Intelligence

Mundane Tasks

1.Perception

•Vision

•Speech

2.Natural Language

•Understanding

•Generation

•Translation

3.Common Sense Reasoning

4.Robot Control

Formal Tasks

1.Games

•Chess

•Backgammons

•Checkers-go

2.Mathematics

•Logic

•Geometric

•Integral calculus

Expert Tasks

1.Engineering

•Design

•Fault Finding

♦Medical Diagnosis

♦Financial Analysis

♦Scientific Analysis

The fields of AI are domains that require specialized expertise without the help of common sense reasoning.

Research into AI shows that intelligence requires knowledge. It is the basic thrust behind every intelligent system. The properties of knowledge are

1.It is voluminous.

2.It is hard to characterize accurately.

3.It is constantly changing.

4.It is well organized and corresponds to the way it will be used.

The AI technique is a method that exploits knowledge that should be represented in such a way that

•Knowledge captures generalization. It is not necessary to represent each individual situation separately. Situations that share important properties are grouped together. Otherwise, a lot of memory and updating would be required.

•It can be understood by the people who must provide it.

•It can be easily modified to correct errors.

•It can be used in many situations even if it is not totally accurate.

1.6Features of AI Programs

a)AI problems have combinatorial explosions of solutions.

b)AI programs manipulate symbolic information to a large extent, in contrast to conventional programs that deal with numeric processing.

c)AI programs use heuristic search techniques to solve problems and prune search trees. One of the techniques for solving problems in artificial intelligence is searching. Searching can be described as solving a problem using a set of states (a situation). A search procedure starts from an initial state and goes through the intermediate states until finally reaching a target state. For example, in solving a puzzle, the initial state is the unsolved puzzle, the intermediate states are the steps taken to solve the puzzle, and the target state is the situation in which the puzzle is solved. The set of all states used by a searching process is referred to as the search space.

d)An AI program must have large quantities of knowledge that must be represented in a form such that a system working on the knowledge can easily manipulate it.

e)AI programs deal with real life problems. AI programs help people make the right decisions.

f)AI programs have the ability to learn.

1.7Importance of AI

a)Organizations that use AI applications become more diverse because these applications provide the ability to analyze data across multiple variables, and can help with fraud detection and customer relationship management. All such things are very important from a competitive point of view.

b)AI is a branch of science that deals with helping machines that find solutions to complex problems in a human-like fashion by borrowing characteristics from human intelligence and applying them as algorithms in a computer-friendly way.

c)AI is generally associated with computer science, but it has its roots in variety of fields, such as math, psychology, cognition, biology, and philosophy. Thus, combining knowledge from all these fields benefits the development of an intelligent artificial being.

d)AI is a machine that can behave like an ordinary human. One of the meanings of the word “perception” is understanding what is received through the senses—sight, hearing, touch, smell, and taste. A human being sees a scene through the eyes, and the brain interprets it to extract the type of objects in the scene. A human hears a set of voice signals through the ears, and the brain interprets it as a meaningful sentence.

1.8What Can Artificial Intelligence Systems Do?

Today’s AI systems have been able to achieve limited success in some of these tasks:

•In computer vision, the systems are capable of facial recognition.

•In robotics, we have been able to make vehicles that are mostly autonomous.

•In natural language processing, we have systems that are capable of simple machine translation.

•Today’s expert systems can carry out medical diagnoses in a narrow domain.

•Speech understanding systems are capable of recognizing several thousand words of continuous speech.

•Learning systems are capable of performing text categorization into about 1,000 topics.

•AI systems can play games at the Grand Master level in chess (world champion) and checkers.

1.9What Can Artificial Intelligence Systems Not Do Yet?

•Understand natural language robustly (e.g., read and understand articles in a newspaper).

•Surf the web.

•Interpret an arbitrary visual scene.

•Learn a natural language.

•Construct plans in dynamic real-time domains.

•Exhibit true autonomy and intelligence.

1.10Advantages of AI

a)AI is used in various areas like diagnosis, medicine, image processing, and game playing; complex tasks in such fields can be performed efficiently and reliably.

b)AI can perform multiple tasks at once, such as tasks that would be too difficult or time consuming when carried out by humans. These tasks include mathematical equations that are used to design and operate video games or autopilots used by airplanes to fly planes in normal situations and aid the crew in emergencies.

c)AI helps in the mass production of industrial parts to make sure parts are accurate and to specifications.

d)AI makes life safer and more pleasurable for people at every stage of modern life.

e)AI machines help in the continuity of work in various fields, as the machines can constantly monitor complex situations.

f)AI can take on stressful and complex work that humans may struggle with or cannot do. Machines have no need for sleep, they don’t get ill, and there is no need for breaks. Doing tasks without getting tired is a significant benefit offered by artificial intelligence. AI can get a specific task finished without a coffee break or lunch break, unlike humans, who require a break. A machine can also complete a particular job almost instantly.

g)AI can replace human beings in some specific jobs at stores and perform some of a household’s day-to-day activities, helping to address manpower problems.

h)AI can help hospitals providing food and medicines where humans may be exposed to disease. AI has its application in variety of fields, such as robotics. Robots can be used in manufacturing industries or other industries for doing tasks that are harmful to humans.

i)AI helps researchers in aeronautics better know the universe.

1.11Disadvantages of Artificial Intelligence

a)It is true that AI has lot of advantages in various fields (such as in chess, where a computer can beat a human). Expert systems assist industry with a wide range of diagnostic software. Robots are used to perform complex and dangerous work. Optical character recognition and speech recognition have advanced enough to have many practical applications. AI has disadvantages, though. For example, without a massive amount of storage, the simultaneous real-time retrieval of multisensory data is out of reach. Natural language processing suffers from this problem. It requires an understanding of language, culture, history, and emotions to be able to translate a sentence. A robot does not have the common sense and the reasoning power to understand a word like “outside.” Common sense requires a large amount of knowledge.

b)The main disadvantage of AI is that it lacks the pattern recognition tools needed to succeed. The study of AI began formally at Dartmouth College in 1956 as an effort by a group of scientists to evaluate and mechanically replicate human intelligence on the assumption that “every aspect of learning or any other feature of intelligence can be so precisely described that a machine can be made to simulate it.” Their objective was to write computer programs, which could finally create human level intelligence in computers and robots. Those early scientists failed to realize that the mind uses pattern recognition and not computation. They also underestimated the memory storage capacity required to achieve such an ambitious objective.

c)Nature has provided a large memory capacity in humans to sustain life on an unimaginable scale. The evolutionary process logically assembled these in cell memories. The DNA of every living thing on the planet has digital, error-correcting, and self-replicating codes. These vast blueprints improved with each generation across millions of years. AI’s disadvantage is that it presently lacks the means to store a comparable size of memory and also lacks a clear strategy for instantly accessing this enormous memory store.

d)Nature has assembled ascending levels of knowledge in the immune system, the spinal cord, the reticular system, the limbic system, and the prefrontal regions. Millions of potentially pathogenic organisms and substances had to be neutralized. “Knowledge” in the spinal cord coordinates the movements of muscles millisecond by millisecond. Memories for myriad smells enabled the reptilian systems to distinguish between prey and predator. “Knowledge” in the limbic system responded suitably to a wide range of events, which trigger anger and fear, or jealousy and despair. The disadvantage of AI is that the computation capabilities of a computer cannot manage the pattern sensing responses of living things, who have assembled this knowledge over millions of years of history and a lifetime of experience, play, and imagination.

e)Human intelligence adds new levels to animal intelligence. The great achievements in science and art are based on the stored knowledge of millions of relationships between numerous fields. A work of art is only possible through an immense number of inherited skills and through practice, training, and experience. AI has barely touched on these complex pattern sensing tasks.

f)The possibility of a breakdown is one of the most infamous disadvantages of artificial intelligence. It is like spending much of your money on a car in order to get from one point to another and then needing to deal with the breakdown of the car shortly after buying it. It is the same way for artificial intelligence: it can easily perform a task, but a malfunction can turn the whole thing into nothing.

g)Aside from the possibility of a breakdown, there is also the possibility of losing your essential information. In some cases, because of the malfunction of specific parts, an artificial mind can fall short in keeping in its memory all the files that it must have. This can also occur with humans. If the person who is responsible for maintaining information and collecting data falls asleep on the job, it is accepted that the failure is that person’s mistake. On the other hand, with an artificial mind, it is not assumed, and this really makes the entire difference. This then becomes an important issue. AI or computer systems must be switched off on a daily basis for maintenance. This could be a restraint to output and efficiency, as well as to the interests and benefits of the company in question.

h)AI fails in the speed of knowledge retrieval. An animal mind stores the equivalent of billions of pages of code. This data is evaluated and acted on within milliseconds. The unconscious processes of your immune system utilize internal code recognition systems to attack a detected invader. The olfactory system, using an inbuilt knowledge of smells, enables an animal to instantly recognize a scent and sense danger.

AI cannot compete in the field of real time information retrieval achieved by animals. To succeed, artificial intelligence requires myriad pattern sensing algorithms and the ability to extract contextual knowledge in real time from a vast amount of coded memories.

Exercises

Q1.What is Artificial Intelligence?

Q2.What are the various areas where AI (Artificial Intelligence) can be used?

Q3.Give an explanation of the difference between a strong AI and weak AI.

Q4.Is intelligence a single thing, so that one can ask the “yes or no” question “Is this machine intelligent or not?”

Q5.Is AI about simulating human intelligence?

Q6.What about other comparisons between human and computer intelligence?

Q7.What is the Turing test?

Q8.What are the task domains of AI?

Q9.What are intelligence and intelligent agents?

Q10.What is the cognitive modeling approach?

CHAPTER 2

APPLICATIONS OF ARTIFICIAL INTELLIGENCE

There are many applications of AI, as it can be used in a variety of fields for solving complex problems. Applications range from military uses (for autonomous control and target identification) to the entertainment industry (for computer games and robotic pets). AI has been used in medical diagnosis, stock trading, robot control, law, remote sensing, scientific discovery, and toys.

The various applications areas are as follows.

2.1Finance

Banks use artificial intelligence systems for organizing operations, investing in stocks, and managing properties. Financial institutions use artificial neural network systems to detect charges or claims outside of the norm and identify these for human investigation.

2.2Hospitals and Medicine

Hospitals use artificial intelligence systems to organize bed schedules, perform staff rotations, and provide medical information. Artificial neural networks are used for medical diagnosis.

Other tasks in medicine that can be performed by artificial intelligence include the following:

a)Artificial intelligence systems are used for analyzing medical images to detect diseases. Such systems help scan digital images, such as those from computed tomography. A typical application is the detection of tumors.

b)Heart sound analysis

2.3Robotics

Robotics is the branch of technology that deals with the design, construction, operation, and application of robots, as well as computer systems for their control, sensory feedback, and information processing. Some robots are used in performing dangerous tasks in the manufacturing industry.

A robot is a mechanical or virtual agent, usually an electro-mechanical machine, that is guided by a computer program or electronic circuitry. A robot performs only those tasks for which it is programmed. However, an intelligent robot has sensors, such as cameras, which allow it to respond to changes in the environment.

2.4Expert Systems

An expert system is a computer program that simulates the judgment and behavior of a human with expert knowledge and experience in a particular field. Expert systems contain a knowledge base of accumulated experience and a set of rules that are applied to each particular situation. Sophisticated expert systems can be enhanced with additions to the knowledge base or to the set of rules.

Expert systems are applications of AI that utilize human expertise. Expert systems are used to solve complex problems with help of the expertise stored in the database in rule form. To design an expert system, we need a knowledge engineer, an individual who studies how human experts make decisions and translates the rules into terms that a computer can understand. Expert systems are also known as knowledge-based systems, knowledge-based expert systems, and rule-based systems. They are considered to be “applied artificial intelligence.” The process of developing with an expert system is knowledge engineering. EMYCIN was one of the first “shells” for an expert system, which was created from the MYCIN medical diagnosis system. A production rule system is a rule engine that uses the rule-based approach to implement an expert system.

Expert systems use knowledge representation languages to perform tasks that normally need human expertise. For example, in medicine, an expert system can be used to narrow down a set of symptoms to a likely subset of causes, a task normally carried out by a doctor.

An expert system is built on predefined knowledge about a field of expertise. An expert system in medicine, for example, is built on the knowledge of a doctor specialized in the field for which the system is built: an expert system is supposed to do the same job as the human expert.

2.5Diagnosis

Diagnosis deals with the development of algorithms and techniques that are able to determine whether the behavior of a system is correct. If the system is not functioning correctly, the algorithm should be able to determine, as accurately as possible, which part of the system is failing, and which kind of fault it is facing.

An example of making a diagnosis is the process a garage mechanic uses with an automobile. The mechanic will first try to detect any abnormal behavior based on the observations of the car and his knowledge of this type of vehicle. If he finds out that the behavior is abnormal, the mechanic will try to refine his diagnosis by using new observations and possibly testing the system until he discovers the faulty component. The mechanic plays an important role in the vehicle’s diagnosis.

2.6Pattern Recognition

Pattern recognition is the process of establishing a close match between new stimuli and a previously stored pattern. Pattern recognition systems are used to classify objects based on their attributes and attribute values.

In pattern recognition, a label is assigned to a given input value. An example of pattern recognition is classification, which attempts to assign each input value to one of a given set of classes (for example, determine whether a given email is “spam” or “non-spam”). Pattern recognition algorithms generally aim to provide a reasonable answer for all possible inputs and to perform the “most likely” matching of the inputs, taking into account their statistical variation.

Pattern recognition is generally categorized according to the type of learning procedure used to generate the output value.

Supervised learning assumes that a set of training data (the training set) has been provided, and it consists of a set of instances that have been properly labeled by hand with the correct output.

Unsupervised learning assumes training data has not been hand-labeled, and attempts to find inherent patterns in the data that can then be used to determine the correct output value for new data instances. A combination of the two is called semi-supervised learning, which uses a combination of labeled and unlabeled data (typically a small set of labeled data combined with a large amount of unlabeled data).

FIGURE 2.1 Pattern Recognition Process

2.7Natural Language Processing

Natural language processing (NLP) is used for analyzing and representing natural text at one or more levels of linguistic analysis to obtain human-like language processing. NLP is related to human–computer interactions.

NLP is a branch of artificial intelligence that deals with analyzing, understanding, and generating the natural languages humans use. One of the challenges in NLP is teaching computers to understand the way humans learn and use language.

For example, consider the sentence “Baby swallows fly.” This simple sentence has multiple meanings, depending on whether the word “swallows” or the word “fly” is used as the verb, which also determines whether “baby” is used as a noun or an adjective. In the case of human communication, the meaning of the sentence depends on the context in which it was communicated. This sentence presents problems for software, which must first be programmed to understand the context and linguistic structures.

Computers can’t understand natural language, so researchers are trying to make them more intelligent. NLP is divided into following subfields:

a)natural language understanding

b)analysis of language to provide meaningful representation

c)natural language generation

d)production of language from representation

Steps in NLP:

a)The first step in natural language processing is speech recognition. In this step, a speech signal is analyzed and the sequence of words it contains is extracted. The input to the speech recognition sub-system is a continuous (analog) signal: the output is a sequence of words. The signal needs to be divided into different sounds, sometimes called phonemes. The sounds then need to be combined into words.

b)The syntactic analysis step is used to define how words are to be grouped in a sentence. This is a difficult task in a language like English, in which the function of a word in a sentence is not determined by its position in the sentence. For example, consider the following two sentences.

Mary rewarded John.

John was rewarded by Mary.

c)It is always John who is rewarded, but in the first sentence, John is in the last position and Mary is in the first position. A machine that hears any of the above sentences needs to interpret them correctly and come to the same conclusion, no matter which sentence is heard.

d)The semantic analysis extracts the meaning of a sentence after it has been syntactically analyzed. This analysis creates a representation of the objects involved in the sentence, their relationships, and their attributes. The analysis can use any of the knowledge representation schemes. For example, the sentence “John has a dog” can be represented using predicate logic.

∃ xdog(x) has (John, x)

The three previous steps—speech recognition, syntax analysis, and semantic analysis—can create a knowledge representation of a spoken sentence. In most cases, another step, pragmatic analysis, is needed to further clarify the purpose of the sentence and remove ambiguities.

2.8Game Playing

In 1960, Arthur Samuel built the first game playing program, which learned from its mistakes and improved its performance. Game playing has great role in AI because of the following:

•Rules are limited and little knowledge is required.

•Games provide structural tasks that are easy to measure as a success or failure.

•Game playing simulates real-life situations.

2.9Image Processing