Artificial Intelligence
Artificial Intelligence
Uploaded by
Santhosh ReddyCopyright:
Available Formats
Artificial Intelligence
Artificial Intelligence
Uploaded by
Santhosh ReddyCopyright
Available Formats
Share this document
Did you find this document useful?
Is this content inappropriate?
Copyright:
Available Formats
Artificial Intelligence
Artificial Intelligence
Uploaded by
Santhosh ReddyCopyright:
Available Formats
ARTIFICIAL INTELLIGENCE
Technical Seminar Report submitted to
Jawaharlal Nehru Technological University Hyderabad
in partial fulfilment for the award of degree of
Bachelor of Technology
in
Computer Science & Engineering(Cyber Security)
by
VANGALA SANTHOSH REDDY
Roll No:20X31A6249
Under the Guidance of
Mrs. M.KARUNA
ASSISTANT PROFESSOR
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING-
(CYBERSECURITY)
SRI INDU INSTITUTE OF ENGINEERING & TECHNOLOGY
(Affiliated to JNTUH, Hyderabad, Approved by AICTE, New Delhi)
Sheriguda (V), Ibrahimpatnam (M), R.R.Dist., Telangana- 501510
SRI INDU INSTITUTE OF ENGINEERING & TECHNOLOGY
(Affiliated to JNTUH, Hyderabad, Approved by AICTE, New Delhi)
Sheriguda (V), Ibrahimpatnam (M), R.R.Dist., Telangana- 501510.
CERTIFICATE
DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING -
(CYBERSECURITY)
This is to certify that the dissertation entitled “Artificial Intelligence”, being
submitted by Vangala Santhosh Reddy, bearing Roll No: 20X31A6249, to Jawaharlal Nehru
TechnologicalUniversityHyderabadinpartialfulfilmentoftherequirementsfortheawardofthe
degree of Bachelor of Technology in Computer Science & Engineering-(Cyber Security), is a
record of bonafide work carried out by him. The results of investigations enclosed in this report
havebeenverifiedandfoundsatisfactory.Theresultsembodiedinthisdissertation havenotbeen
submitted to any other University or Institute for the award of any other degree.
INTERNAL GUIDE HEAD OF THE DEPARTMENT
SRI INDU INSTITUTE OF ENGINEERING & TECHNOLOGY
(Affiliated to JNTUH, Hyderabad, Approved by AICTE, New Delhi)
Sheriguda (V), Ibrahimpatnam (M), R.R.Dist., Telangana- 501510.
DECLARATION
I,VANGALASANTHOSHREDDYbearingRollNo20X31A6249, herebycertify
that the dissertation entitled “ARTIFICIAL INTELLIGENCE”, carried out under the
guidance of Mrs.M.KARUNA ,ASSISTANT PROFESSOR is submitted to Jawaharlal
Nehru Technological University Hyderabad in partial fulfilmentoftherequirementsfor
the award of the degree of Bachelor of Technology in Computer Science &
Engineering-(CyberSecurity).Thisisarecordofbonafideworkcarriedoutbymeandthe
results embodied in this dissertation have notbeenreproducedorcopiedfromanysource.
TheresultsembodiedinthisdissertationhavenotbeensubmittedtoanyotherUniversityor
Institute for the award of any other degree.
Date: V Santhosh Reddy
Roll No:20X31A6249
Department of CSE,SIIET
ACKNOWLEDGEMENT
WithgreatpleasureItakethisopportunitytoexpressmyheartfeltgratitudetoallthepersons
who helped me in making this technical seminar successful.
First of all, I express my sincere thanks to Mr. R. VENKAT RAO, Chairman, SriIndu
Group of Institutions, for his continuous encouragement.
IamhighlyindebtedtoPrincipal,Dr.I.SATYANARAYANAforgivingmethepermission
to carry out this work.
IwouldliketothankDr.B.G.ObulaReddyProfessor&HeadoftheDepartment(CSE),
for giving support throughout the period of my study in SIIET. I am grateful for his valuable
suggestions and guidance during the execution of this work.
My sincere thanks to technical seminar guide Mrs.M.Karuna,Assistant Professor for
potentially explaining the entire system and clarifying the queries at every stage of the work.
My whole hearted thanks to the staff of Computer Science And Engineering(Cyber
Security)who cooperated with us for the completionof the work in time.
I also thank my parents and friends who aided me in completion of the Technical seminar.
SANTHOSH REDDY
V
Roll No: 20X31A6249
i
ABSTRACT
ArtificialIntelligence(A.I.)isamultidisciplinaryfieldwhosegoalistoautomateactivities
that presently require human intelligence. Recent successes in A.I. include computerisedmedical
diagnosticians and systems that automaticallycustomisehardwaretoparticularuserrequirements.
The major problem areas addressed in A.I. can be summarised as Perception, Manipulation,
Reasoning, Communication, and Learning. Perception is concerned with building models of the
physicalworldfromsensoryinput(visual,audio,etc.).Manipulationisconcernedwitharticulating
appendages (e.g., mechanical arms, locomotion devices) in order to effect a desired state in the
physical world. Reasoning is concerned with higher level cognitive functions such as planning,
drawing inferential conclusions from a world model, diagnosing, designing, etc. Communication
treats the problem of understanding and conveying information through the use of language.
Finally, Learning treats the problem of automatically improving system performance over time
based on the system's experience. Many important technical concepts have arisen from A.I. that
unify these diverse problem areas and that form the foundation of the scientific discipline.
Generally,A.I.systemsfunctionbasedonaKnowledgeBaseoffactsandrulesthatcharacterisethe
system'sdomainofproficiency.TheelementsofaKnowledgeBaseconsistofindependentlyvalid
(or at least plausible) chunks of information. The system must automatically organise and utilise
thisinformationtosolvethespecificproblemsthatitencounters.Thisorganisationprocesscanbe
generally characterised as a Search directed toward specific goals. The search is made complex
because of the need to determine the relevance of information and because of the frequent
occurrenceofuncertainandambiguousdata.HeuristicsprovidetheA.I.systemwithamechanism
for focusing its attention and controlling its searching processes.
ii
CONTENTS
TOPICS PAGE NO
ACKNOWLEDGEMENT i
ABSTRACT ii
CONTENTS iii
LIST OF FIGURES v
CHAPTER-1: INTRODUCTION 1
1.1 Classical definitions 1
1.2 History of AI 1
1.3 Intelligent Agent 2
1.4 Topics in AI 2
1.5 Types of AI 2
1.6 Future of AI 4
1.7 Artificial Intelligence job opportunities 5
CHAPTER-2:WORKING MODEL 6
2.1 Machine Learning 6
2.2 The Different Types of Machine Learning 7
2.3 The Big Issue in Training Machine Learning Models 7
2.4 Mostly Used AI Models 8
iii
CHAPTER-3:ARCHITECTURE 11
3.1 Signal processing 11
3.2 SpeechRecognition 11
3.3 Symbolic reasoning 11
3.4 Computer vision 12
3.5 Image processing 13
3.6 speech recognition 13
3.7 Pattern recognition 14
3.8 Neural Network Architectures 14
3.19 Deep Learning 15
3.10 Reinforcement Learning Architectures 16
3.11 Basic Information Processing 17
CHAPTER-4:APPLICATIONS 17
4.1 AI Application in E Commerce 17
4.2 Applications Of Artificial Intelligence in Education 17
4.3. Applications of Artificial Intelligence in Lifestyle 19
4.4 Applications of Artificial in Intelligence in Navigation 19
4.5 Applications of Artificial Intelligence in Robotics 20
4.6 Applications ofArtificial Intelligence in HumanResource 21
4.7 Applications of Artificial Intelligence in Healthcare 21
CHAPTER-5:.CONCLUSION 23
CHAPTER-6:REFERENCES 25
iv
LIST OF FIGURES
PAGE NO
Fig -1.1 INTELLIGENT AGENT 2
Fig -1.2 TOPICS IN AI 2
Fig -1.3 FUZZY LOGIC SYSTEMS 3
Fig -2.1 NEURAL NETWORKS 8
Fig -2.2 DECISION TREES 9
Fig -2.3 RANDOM FOREST 10
Fig -3.1 ARCHITECTURE 11
ig -3.2 INFORMATION
F
17
PROCESSING
v
1.INTRODUCTION
The fundamental goal of AI research 1.2 History of AI :-
is not merely to mimicIntelligenceorsome
beganinantiquity,withmyths,stories
clever fake not at all,AI can be simply
andrumoursofartificialbeingsendowedwith
expressedastheMachineswithMindsinfull
intelligence or consciousness by master
and literal sense.
craftsmen. The seeds of modern AI were
Artificialintelligenceisthesimulation planted by philosophers who attempted to
ofhumanintelligenceprocessesbymachines, describetheprocessofhumanthinkingasthe
especially computer systems.Artificial mechanical manipulation of symbols. This
intelligence is the simulation of human work culminated in the invention of the
intelligenceprocessesbymachines,especially programmable digital computer inthe1940s,
computer systems a machine based on the abstract essence of
1.1 Some Classical definitions:- mathematical reasoning. This device and the
ideasbehinditinspiredahandfulofscientists
We call programs intelligent if they
tobeginseriouslydiscussingthepossibilityof
exhibit behaviours that wouldberegardedas
building an electronic brain.
intelligent if they were exhibited by human
beings.-Herbert Simon The field of AIresearchwasfounded
at a workshop held on the campus of
Physicists ask what kind of place this
Dartmouth College, USA during thesummer
universe is and seek to characterise its
of 1956. Those who attended would become
behaviour systematically.Biologistsaskwhat
the leaders of AIresearchfordecades.Many
it means for a physical system to be living.
ofthempredictedthatamachineasintelligent
We in Al wonder what kind of
asahumanbeingwouldexistinnomorethan
information-processing system can ask such
ageneration,andtheyweregivenmillionsof
questions.- Avron Barr and Edward
dollars to make this vision come true
Feigenbaum
Eventually, it became obvious that
Al is the study of techniques for solving
commercial developers and researchers had
exponentially hard problems in polynomial
grossly underestimated the difficulty of the
time by exploiting knowledge about the
project.[3] In1974,inresponsetothecriticism
problem domain.- Elaine Rich
from James Lighthill and ongoing pressure
Alisthestudyofmentalfacultiesthroughthe from congress, the U.S. and British
use of computational models. - Eugene Governments stopped funding undirected
Charniak and Drew McDermot research into artificial intelligence, and the
difficult years that followed would later be
knownasan"A
Iwinter".Sevenyearslater,a
1
visionary initiative by the Japanese
Government inspired governments and
industrytoprovideAIwithbillionsofdollars,
but by the late 1980s the investors became
disillusioned and withdrew funding again.
Investment andinterestinAIboomed
in the first decades of the 21stcenturywhen
machine learning wassuccessfullyappliedto
manyproblemsinacademiaandindustrydue
to new methods, the application of powerful
computer hardware, and the collection of Fig 1 Intelligent Agent
immense data sets intelligent agent in a world carries a
model of the world in its "head". Themodel
How Al Got Its Name?
may be an abstraction. A self aware agent
The name artificial intelligence is
would model itself in the world model. Dee
credited to John McCarthy who, along with
awareness may require that the agent
Marvin Minsky and Claude Shannon
represent(beawareof)itselfmodellingtheW
(1916-2001), organised the Dartmouth
(From A First Course in Al - Deepak
Conference in 1956.
Khemani
The conference was to be a "two
1.4 Topics in AI:
month,ten-manstudyofartificialintelligence
... on the basis of the conjecture that every Fig:2 topics in AI
aspect of learning or any other feature of
intelligence can in principle be so precisely
described that a machine can be made to
simulate it".
1.3 Intelligent Agent
An AI system can be defined as the
study of the rational agent and its
environment. The agents sense the
environment through sensorsandactontheir
environment through actuators. An AI agent
can have mental properties such as
knowledge, belief, intention, etc.
2
1.4.1 Expert Systems approximation techniques from neural
networks.
Examples − Flight-tracking systems,
Clinical systems. Examples − Consumer electronics,
automobiles, etc
1.4.2 Natural Language Processing
Natural Language Processing is a
subfield ofArtificialIntelligencerelatedwith
Fig 1.3
the interactions between computers and
human language. Apart from above, also qualitative
reasoning,speech synthesis,handling
Examples: Google Now feature,
uncertainty,image processing,knowledge
speech recognition, Automatic voice output.
discovery,tactile sensing,robot
1.4.3 Neural Networks
controls,problem solving and plannings are
A neural network is a method in field or domains in AI
artificial intelligence that teaches computers
toprocessdatainawaythatisinspiredbythe 1.4.6 Domains of study:-
human brain. AI has becoming increasingly popular there
Examples − Pattern recognition are numerousdomainsorfieldstolookat,or
systems such as face recognition, character requires a vast research for further
recognition, handwriting recognition. advancement below are some of those fields
where AI is used or needed to be utilised
1.4.4 Robotics
robotics,design,construction,anduse 1.4.7 Astronomy
of machines (robots) to perform tasks done Artificial Intelligence can be very
traditionally by human beings. useful to solve complex universe problems.
Examples − Industrial robots for AI technology can be helpful for
moving, spraying, painting, precision understanding the universe such as how it
checking, drilling, cleaning,coating,carving, works, origin, etc.
etc. Healthcare
1.4.5 Fuzzy Logic Systems AIhasbecomemoreadvantageousfor
A fuzzy neural network or thehealthcareindustryandisgoingtohavea
neuro-fuzzysystemisalearningmachinethat significant impact on this industry.
finds the parameters of a fuzzy system (i.e., Healthcare Industries are applying AI
fuzzy sets, fuzzy rules) by exploiting to make a better and faster diagnosis than
3
humans. AI can help doctors with diagnoses agriculture robotics, solid and crop
and can inform when patients are worsening monitoring, predictive analysis. AI in
so that medical help can reach the patient agriculture can be very helpful for farmers
before hospitalisation.
1.4.11 E-commerce
1.4.8 Automotive Industry AI is providing acompetitiveedgeto
SomeAutomotiveindustriesareusing the e-commerce industry, and it is becoming
AI to provide virtual assistants to theirusers moredemandinginthee-commercebusiness.
for better performance. Such as Tesla has AI is helpingshopperstodiscoverassociated
introduced TeslaBot, an intelligent virtual products with recommended size, colour, or
assistant. even brands
Various Industries are currently
workingondevelopingself-drivencarswhich
1.5 Types of AI:-
can make your journey more safe and secure.
1.5.1 Narrow AI:
1.4.9 Robotics:
Artificial Intelligence has a It is capable of completing dedicated tasks
remarkable role inRobotics.Usually,general with intelligence. The current stage of AI is
robots are programmed such that they can narrow AI.
perform some repetitive task, but with the
1.5.2 General AI:
help of AI, we can create intelligent robots
which can perform tasks with their own Artificial General IntelligenceorAGI
experiences without being pre-programmed. defines the machines that can show human
Humanoid Robots are the best intelligence.
examples for AI in robotics, recently the
1.5.3 Super AI:
intelligent Humanoid robot named Erica and
Sophia has been developed which can talk Super AI refers toself-awareAIwith
and behave like humans. cognitiveabilitiesthatsurpassthatofhumans.
It is a level where machines candoanytask
1.4.10 Agriculture
that a human can do with cognitive properties.
Agriculture is an area which requires
variousresources,labour,money,andtimefor 1.6 Future of AI:-
best results. Nowadays agriculture is
Undoubtedly, Artificial Intelligence
becoming digital, and AI is emerging in this
(AI) is a revolutionary field of computer
field. Agriculture is applying AI as
4
science, which is ready to become the main 1.7 Artificial Intelligence job
component of various emerging technologies opportunities
like big data, robotics, and IoT. It will
According to a report published by
continuetoactasatechnologicalinnovatorin
Forbes,AIjobopportunitiesarecontinuously
the coming years. Injustafewyears,AIhas
increasing at 74% annually.Itisano-brainer
become a realityfromfantasy.Machinesthat
that today, AI is one of the most in-demand
help humans with intelligence are not justin
technologies, and it has an impact in almost
sci-fi movies but also in the real world. At
every field. As a result, the demand for AI
this time, we live in a world of Artificial
professionals continues to grow. As the
Intelligence that was just a story though for
numberofjobopportunitiesincreases,itisthe
some years.
best time to explore your career in A
We are using AI technology in our
daily lives either unknowingly orknowingly,
and somewhere it has become a part of our
life. Ranging from Alexa/Siri to Chatbots,
everyoneiscarryingAIintheirdailyroutine.
The development and evolution of this
technology are happening at a rapid pace.
However, itwasnotassmoothandeasyasit
seemed to us. It has taken several years and
lots of hard work & contributions ofvarious
people to take AI at this stage. Beingsucha
revolutionary technology, AI also deals with
many controversies about its future and
impact on Human beings. It may be
dangerous, but also a great opportunity. AI
will be deployed to enhance both defensive
and offensive cyber operations. Additionally,
newmeansofcyber-attackwillbeinventedto
take advantageofparticularvulnerabilitiesof
AI technology
5
2. WORKING MODEL
Artificial Intelligence is one of the should be aware of a critical difference
most fascinating fields of computer science. between the two.
Artificial Intelligence studies aim to create a
As we previously defined, Artificial
machinethatcanreplicatehumanintelligence
Intelligence is concerned with creating
in real time. True AI that can think like a
machines that simulate human thoughts,
human has not been achieved. However,this
intelligence, and behaviours.
doesn’t mean we can’tbenefitfromusingAI
algorithms. On the other hand,MachineLearning
is concerned with providing machines with
Today, we use AI models for several
the ability to learn for themselves from
analytical and decision-making tasks. An AI
experienceandlessonswithouttheneedtobe
modelisaprogramoralgorithmthatrelieson
explicitly programmed.
training data to recognize patterns and make
predictionsordecisions.Themoredatapoints All Machine Learning models are AI
an AI model receives, the more accurate it models,butnotallAImodelswillnecessarily
can be in its data analysis and forecasts. be Machine Learning models. This is an
important distinction to make and will help
AI models rely on computer vision,
you understand this topic ingreaterdetail.A
natural language processing, and Machine
Machine Learning model is an essential part
Learning to recognize different patterns. AI
of.Afterall,abigpartofhumanintelligence
models also use decision-making algorithms
is learning things and predicting future
tolearnfromtheirtraining,collectandreview
outcomes based on past experiences and
data points, and ultimately apply their
lessons. Similarly, AI models learn in their
learning to achieve their predefined goals.
learning phase based
AI models are very good at solving
All Machine Learning models are AI
complex problems with a large amount of
models,butnotallAImodelswillnecessarily
data. As a result, they can accurately solve
be Machine Learning models. This is an
complexproblemswithaveryhighdegreeof
important distinction to make and will help
accuracy
you understand this topic in greater detail.
2.1 Machine Learning
A Machine Learning model is an
Many people erroneously conflate Machine essential part of . After all, a big part of
Learning and Artificial Intelligence. This is human intelligence is learning things and
likely because Machine Learning is a subset predicting future outcomes based on past
of Artificial Intelligence. However, you experiencesandlessons.Similarly,AImodels
6
learn in their learning phase based on Unsupervised Machine Learning
annotated data.
AImodelsdon’thavetobedeveloped
2.2 The Different Types of Machine through human training. Instead, in an
Learning unsupervised learning model, software trains
the algorithm. In some cases, the training
TherearevarioustypesofAImodels,
method used by the training software will
and we will go over some of the most
mimic that of a human, but they don’t
prominent ones later in this post, but before
necessarily have to teach in the same way.
we do that, we must go over the different
types of Machine Learning. AI models built with unsupervised
learningareoftenusedtoperformdescriptive
TherearethreemaintypesofMachine
analysis. These tasks include content
Learning:
summary,classification,extraction,andvideo
● Supervised analytics. These types of AI models can
● Unsupervised identify patterns and categorise data without
● Semi-supervised human training.
Semi-Supervised Machine Learning
Supervised Machine Learning
Semi-supervised learning models
In a supervised learning model, a
combine a little bit of both previous models
humantrainsorteachesthealgorithmwhatto
we have discussed. In a semi-supervised
look for. Often thispersondoingthetraining
learning setting, a human does part of the
is a data scientist who is a subject matter
training,andsoftwareislefttohandletherest
expertinwhatevertaskthealgorithmisbeing
based on the initial training done by the
taught to do.
human.
Forexample,analgorithmworkingon
Since AI models built with
image recognition might be taught by a data
semi-supervised learning get a bit of both
scientistinwhichimagescontainoffensiveor
learning styles, they can also perform both
explicit content. Thealgorithmthenusesthis
predictive and descriptive analysis tasks
information to findotherimageswithsimilar
depending on the purpose they are designed
content.
for.
AI models built with supervised
learning, like support vector machines, are 2.3 The Big Issue in Training
often used to perform predictive analysis. Machine Learning Models
These models use past decisions made by
Bias is the biggesttrainingissuewith
subject matter experts to predict future
MLmodels.Thechallengefordevelopersand
choices that an expert might make.
data scientists is to try and reduce training
7
biastonear-zero.Completelyeliminatingbias Deep Neural Networks
might be impossible, but reducing bias as
Thedeepneuralnetworkisoneofthe
much as possible is critical.
most popular AI/ML models. The design for
Training bias can occur in all ML this deep learning model wasinspiredbythe
models,eveninunsupervisedsettings.Sincea human brain and its neural network.ThisAI
ML model is designed to perform specific model uses layers of artificial neurons to
tasks, the last thing researchers and data combine multiple inputsandprovideasingle
scientists want is bias. A famousexampleof output value. Hence the name, deep learning.
model bias came from Amazon.
Deep neuralnetworkshavebeenused
AmazondevelopedaMLmodeltovet widelyinmobileappdevelopmenttoprovide
applicants for development positions and image and speech recognition services and
highlight the most impressive resumes for naturallanguageprocessing.Neuralnetworks
human review. Amazon gave the model the also help power computervision applications.
resumes of the best developers currently
This AI model represents the cutting
working in the field for training data.
edgeofArtificialIntelligence.Itisveryadept
The problem? Since the majority of at solving complex problems that possess
the development industry was male at the large data sets.
time, the model devalued and even
Deeplearningneuralnetworkswillbe
discriminated against women applicants.
instrumental in achieving the true computer
This is an example of unintentional visionandAIstandardsthatweassociatewith
biasbeinglearnedbythemodel,anditisvery human intelligence and science fiction stories.
importantfordevelopersanddatascientiststo
root these biases out before they have
real-world consequences.
2.4 Mostly Used AI Models
There are several different AI models, and
they all work a little bit differently. Someof
themostpopularmodelsyoumightfindinan
AI model library include:
fig:-3 Neural networks
● Deep neural networks
The architecture of a neural network,
● Linear regression
each layer consists of nodes. The number of
● Logistic regression hidden layers is optional. Source:MathWorks
● Decision trees
● Random forest
8
Are Artificial Neural NetworksAble model because it is only used to solve
to Outperform the Human Brain? classification-based problems.
Logistic regression is the best AI
The world’s most sophisticated
model for solving a binary classification
artificial neural networks cannot compete
problem.Thismodelisadeptatpredictingthe
with the brain of a human being. In fact,
valueorclassofadependentdatapointbased
scientistsandneuralresearcherscan’texplain
on a set of independent variables.
how exactly the brain works to form
conscious thought and complete complex
Decision Trees
tasks.
This AI model is straightforward and
The hardware metaphor is often used
also highly efficient. The decision tree uses
to describe how the brain functions, but to
availabledatafrompastdecisionstoarriveat
this point, this comparison does little to
aconclusion.Thesetreesoftenfollowabasic
broaden the way we understand our own
if/then pattern. For example, if you eat a
minds
sandwich at home, then you willnotneedto
Linear Regression buy lunch.
Decision trees can be used to solve
This AI model is very popular with
both regression and classification problems.
data scientists working in statistics. Linear
In addition, rudimentary decision trees
regression is based on a supervised learning
powered the earliest forms of predictive
model. These AI models are tasked with
analytics.
identifyingtherelationshipbetweeninputand
output variables.
A linear regression modelcanpredict
thevalueofadependentvariablebasedonthe
value of an independent variable. These fig:4 decision tree
models are used in linear discriminant
analysis for several industries, including
healthcare, insurance, eCommerce, and
banking.
Logistic Regression
ThisisanotherpopularAImodel,and
it is closely related to the linear regression
model. However, the logistic regression
model is different from the linear regression
9
Random Forest
If one decision tree is a powerful AI
model, how mighty is an entire forest? A
random forest is a collection of multiple
decision trees.
Each decision tree returns its own
result or decision,whichisthenmergedwith
theresultsfromeveryothertreeintheforest.
Finally, the combined results make a more
accurate final prediction or decision. The
random forest is a great AI modelwhenyou
have a large data set. This model isusedfor
solving both regression and classification
problems. Modern predictive analytics are
powered in large part by random forest
models.
fig:5 random forest
10
3. ARCHITECTURE
3.3 Symbolic reasoning
Symbolic reasoning is a fundamental
concept in artificial intelligence (AI) that
involves manipulating symbols, logic, and
abstract representations to perform various
cognitive tasks. Unlike some other AI
approaches that rely on numerical
computations and statistical methods,
fig:6 Architecture symbolic reasoning focuses on using
symbolic information to represent and
The term "architecture of AI" can
manipulate knowledge. Here are some key
refer to different aspects of artificial
aspects of symbolic reasoning in AI:
intelligence, including both the design of AI
systems and the underlying computational 1. Symbol Representation: In
modelsthatpowerthem.Here,I'llprovidean symbolic reasoning, information is
overview of a few key aspects represented using symbols, which can be
abstract entities representing concepts,
3.1 Signal processing
objects, relationships, and more. These
Signal processing plays a crucial role symbols can be manipulated using logical
in various aspects of artificial intelligence rules and operations.
(AI). It involves the manipulation, analysis,
2. Logic and Inference: Symbolic
and interpretation of signals, which can be
reasoning heavily relies on formal logic to
any form of data that varies over time or
perform inference. Logical rules are used to
space. In the context of AI, signals can
derive conclusions from given premises. For
include audio,speech,images,videos,sensor
example, if you know that "all humans are
data,andmore.Herearesomewaysinwhich
mortal" and "Socrates is ahuman,"symbolic
signal processing is integrated with AI:
reasoning can deduce that "Socrates is
3.2 SpeechRecognition mortal."
Image Processing,,Natural Language 3. Knowledge Representation:
processing any activity involving processing Symbolic reasoning is particularly suited for
or manipulating or identifying the input will representing and reasoning about structured
be classified as signal processing knowledge. It allowsAIsystemstorepresent
complex relationships, hierarchies, and rules
11
in a human-readable and interpretable knowledge,capturingnuancedandcontextual
manner. understanding remains a challenge.
4.ExpertSystems:Expertsystemsare 9. Limitations: Symbolic reasoning
a classic example of symbolic reasoning in can struggle with handling uncertainty and
AI. These systems use a knowledge base of dealing with large amounts of unstructured
facts and rules to make decisions or solve data.Itmayalsohavelimitationsincapturing
problems in a specific domain. They rely on the subtleties of human language and
symbolicmanipulationtodrawinferencesand real-world contexts.
provide expert-level advice. 10. Hybrid Approaches: Many AI
5. Semantic Networks and Frames: systems use a combination of symbolic
Semantic networks and frames are graphical reasoning and other techniques, such as
representations that organise knowledge in a machinelearning,toleveragethestrengthsof
way that shows relationships between bothapproaches.ThisisknownashybridAI,
concepts. This enables symbolic reasoning and it aims to overcome the limitations of
systems to navigate and make deductions each approach individually.
based on these relationships. Insummary,symbolicreasoninginAI
6. Planning and Problem Solving: focuses on the manipulation of abstract
Symbolic reasoning is used in planning and symbols and logical rules to represent and
problem-solving tasks. AI systems can reason about knowledge. It has been used in
representaproblemdomainsymbolicallyand variousAIapplications,particularlyinexpert
then use symbolic logic toderiveasequence systems, knowledge representation, and
of actions to achieve a desired goal. problem-solving domains. However, its
7. Natural Language Processing: limitations in handling uncertainty and the
Symbolic reasoning is employed in some complexities of real-world data have led to
natural language processing (NLP) tasks, the development of hybrid approaches that
where symbolic rules and grammatical combine symbolic reasoning with other AI
structures are used to analyse and generate techniques
sentences.Thiscanincludetaskslikeparsing 3.4 Computer vision
sentences, generating responses, and
tasks include methods for acquiring,
understanding meaning.
processing, analysing and understanding
8. Common Sense Reasoning: digital images, and extraction of
Symbolicreasoningischallengedbytheneed high-dimensional data from therealworldin
for systems to possess common-sense order to produce numerical or symbolic
knowledge and reasoning abilities. While information, e.g. in the forms of
symbolic AI can represent explicit decisions.[1][2][3][4] Understanding in this
context means the transformation of visual
12
images (the input to the retina in the human
analog) into descriptions of the world that
make sense to thought processes and can 3.6 Speech recognition
elicit appropriate action. This image
Speech recognition, orspeech-to-text,
understanding can be seen as the
is the ability of a machine or program to
disentangling of symbolic information from
identifywordsspokenaloudandconvertthem
imagedatausingmodelsconstructedwiththe
into readable text. Rudimentary speech
aid of geometry, physics, statistics, and
recognitionsoftwarehasalimitedvocabulary
learning theory
and may only identify words and phrases
3.5 Image processing when spoken clearly. More sophisticated
Image processing is a method to software can handle natural speech, different
perform some operations on an image, in accents and various languages.
order to get an enhanced image or toextract Speechrecognitionusesabroadarray
someusefulinformationfromit.Itisatypeof of research in computer science, linguistics
signal processing in which input is animage and computer engineering. Many modern
and output may be image or devices and text-focused programs have
characteristics/features associated with that speechrecognitionfunctionsinthemtoallow
image.Nowadays,imageprocessingisamong for easier or hands-free use of a device.
rapidly growingtechnologies.Itformsacore Speech recognition and voice
research area within engineering and recognitionaretwodifferenttechnologiesand
computer science disciplines too. should not be confused:
Image processing basically includes ● Speech recognition isusedtoidentify
the following three steps: words in spoken language.
● Importing the image via image ● Voice recognition is a biometric
acquisition tools; technology for identifying an
● Analysing and manipulating the individual's voice.
image;
● Output in which the result can be How does speech recognition work?
altered image or report that is based
Speech recognition systems use
on image analysis.
computer algorithms to process and interpret
spoken words and convert them into text. A
software program turns the sound a
microphonerecordsintowrittenlanguagethat
computers and humans can understand,
following these four steps: analyse theaudio
13
then break it into parts then digitise it intoa diagnosis.
computer-readable format; anduse an In a typical pattern recognition application,
algorithmtomatchittothemostsuitabletext therawdataisprocessedandconvertedintoa
representation. form that is amenable for a machine to use.
Speechrecognitionsoftwaremustadapttothe Patternrecognitioninvolvestheclassification
highly variableandcontext-specificnatureof and cluster of patterns.
human speech. The software algorithms that In classification, an appropriate classlabelis
process and organise audio into text are assigned to a pattern based onanabstraction
trained ondifferentspeechpatterns,speaking that is generated using a set of training
styles, languages, dialects, accents and patterns or domainknowledge.Classification
phrasings.Thesoftwarealsoseparatesspoken is used in supervised learning.
audio from background noise that often
Clustering generates a partition of the data
accompanies the signal.
which helps decision making, the specific
To meet these requirements, speech decision-making activity of interest to us.
recognition systems use two types of models: Clustering is used in unsupervised learning.
Acoustic models. These represent the 3.8 Neural Network Architectures:
relationshipbetweenlinguisticunitsofspeech
Neural networks are an efficient way
and audio signals.
to solve machine learning problems and can
Language models. Here, sounds are matched beusedinvarioussituations.Neuralnetworks
with word sequences to distinguish between offer precision and accuracy. Finding the
words that sound similar. correct neural network for each project can
increase efficiency.
3.7 Pattern recognition
Standard neural networks
theprocessofrecognizingpatternsby
using a machine learning algorithm. Pattern Perceptron - A neural network that
recognition can be defined as the applies a mathematical operation to an input
classification of data based on knowledge value, providing an output variable.
already gained or on statistical information
Feed-Forward Networks - A
extracted from patterns and/or their
multi-layered neural network where the
representation. One of the important aspects
information moves from left to right, or in
of pattern recognition is its application
otherwords,inaforwarddirection.Theinput
potential.
values pass through a series ofhiddenlayers
Examples: Speech recognition, on their way to the output layer.
speaker identification, multimedia document
recognition (MDR), automatic medical
14
● ResidualNetworks(ResNet)-Adeep
feed-forward network with hundreds
of layers. Generative adversarial networks
Generative adversarialnetworks(GAN)area
Recurrent neural networks
type of unsupervised learning where data is
Recurrent neural networks (RNNs) generated from patterns thatwerediscovered
remember previously learned predictions to from the input data. GANs have two main
help make future predictions with accuracy. parts that compete against one another:
Long short term memory network
Generator - creates synthetic data
(LSTM) - LSTM adds extra structures, or
from the learning phase of themodel.Itwill
gates, to an RNN to improve memory
take random datasets and generate a
capabilities.
transformed image.
Echostatenetwork(ESN)-Atypeof
Discriminator-decideswhetherornot
RNN hidden layers that are sparsely
the images produced are fake or genuine.
connected.
GANs are used to help predict what
Convolutional neural networks the next frame in a video might be, text to
Convolutional neural networks image generation, or image to image
(CNNs) are a type of feed-forward network translation.
thatareusedforimageanalysisandlanguage
Transformer neural networks
processing. There are hidden convolutional
layers that form ConvNets and detect Unlike RNNs, transformer neural
patterns. CNNs use features such as edges, networks do not have a concept of
shapes, and textures to detect patterns. timestamps. This enables them to pass
Examples of CNNs include: throughmultipleinputsatonce,makingthem
AlexNet - Contains multiple a more efficient way to process data.
convolutional layers designed for image 3.9 Deep Learning:
recognition.
Deep learning is a branch of machine learning
Visualgeometrygroup(VGG)-VGG which is based on artificial neural networks. It
is similar to AlexNet, but hasmorelayersof is capable of learning complex patterns and
narrow convolutions. relationships within data. In deep learning, we
don’t need to explicitly program everything. It
Capsule networks - Contain nested has become increasingly popular in recent
capsules(groupsofneurons)tocreateamore years due to the advances in processing power
powerful CNN. and the availability of large datasets. Because
15
it is based on artificial neural networks However,availability of cloud computing and
(ANNs) also known as deep neural networks the development of specialised hardware,
(DNNs). These neural networks are inspired such as Graphics Processing Units (GPUs),
by the structure and function of the human has made it easier to train deep neural
brain’s biological neurons, and they are networks.
designed to learn from large amounts of data.
3.10 Reinforcement Learning
Deep Learning is a subfield of Architectures:
Machine Learning that involves the use of
ForAIsystemsthatlearnthroughtrial
neural networks to model and solve complex
and error, like game-playing agents,
problems. Neural networks are modelled after
reinforcementlearningarchitecturesareused.
the structure and function of the human brain
These architectures involve an agent
and consist of layers of interconnected nodes
interacting with an environment tomaximise
that process and transform data.
a reward signal, learning optimal strategies
The key characteristic of Deep Learning is the over time.
use of deep neural networks, which have
Generative Models:
multiple layers of interconnected nodes.
These networks can learn complex These architectures focus on
representations of data by discovering generatingnewdatasamples,suchasimages,
hierarchical patterns and features in the data. music, or text. Variational Autoencoders
Deep Learning algorithms can automatically (VAEs)andGenerativeAdversarialNetworks
learn and improve from data without the need (GANs) are examples of generative models.
for manual feature engineering.
Memory Networks:
Deep Learning has achieved significant These architectures incorporate memory
success in various fields, including image components,allowingAIsystemstostoreand
recognition, natural language processing, retrieve information over time. Memory
speech recognition, and recommendation networks are crucial for tasks requiring
systems. Some of the popular Deep Learning context and reasoning.
architectures include Convolutional Neural
Networks (CNNs), Recurrent Neural Ensemble Methods:
Networks (RNNs), and Deep Belief Networks Sometimes, multiple AI models are
(DBNs). combined to create an ensemble, whereeach
model contributes its predictions. This often
Training deep neural networks typically
improves overall performance and robustness.
requires a large amount of data and
computational resources.
16
Hybrid Architectures: AI systemscan decisions. They help users understand whya
combinedifferentarchitecturestoharnessthe certain decision was made by the AI system.
strengths of each. For instance, a system Remember that the field of AI is
might use aconvolutionalneuralnetworkfor rapidly evolving, and new architectures are
feature extraction followed by a recurrent being developed frequently. The choice of
neural network for sequence modelling. architecturedependsonthespecifictask,data
Attention Mechanisms: availability, computational resources, and
desired outcomes.
Attentionmechanisms,popularisedby
transformer models, allow AI systems to 3.11 Basic Information Processing
focus on specific parts of input data when
making predictions. This has greatly
improved the effectiveness of models in
understanding context.
Meta-learning Architectures:
These architectures focus onenabling
AI systems to learn how to learn. They're
designed to adapt quickly to new tasks with
minimal data.
Fig:7 parts of Ai
Explainable AI (XAI) Architectures:
These architectures aim to provide
transparency and interpretability to AI
17
4. APPLICATIONS
4.1 AI Application in E Commerce 4.2 Applications Of Artificial
Personalized Shopping Intelligence in Education
Artificial Intelligence technology is
Although the education sector is the
used to create recommendation engines
one most influenced by humans, Artificial
through which you can engage better with
Intelligencehasslowlybeguntoseepitsroots
your customers. These recommendations are
into the educationsectoraswell.Eveninthe
made in accordance with their browsing
education sector, this slow transition of
history, preference, and interests. It helps in
Artificial Intelligence has helped increase
improving your relationship with your
productivityamongfacultiesandhelpedthem
customers and their loyalty towards your
concentrate more on students than office or
brand.
administration work.
Some of these applications in this
AI-Powered Assistants
sector include:
Virtual shopping assistants and
chatbots help improve the user experience Administrative Tasks Automated to
while shopping online. Natural Language Aid Educators
Processing is used to make the conversation
Artificial Intelligence can help
sound as human and personal as possible.
educators with non-educational tasks like
Moreover, these assistantscanhavereal-time
task-related duties like facilitating and
engagement with your customers. Did you
automating personalised messages to
know that on amazon.com, soon, customer
students, back-office tasks like grading
service could be handled by chatbots?
paperwork, arranging and facilitating parent
and guardian interactions, routine issue
Fraud Prevention
feedback facilitating, managing enrollment,
Credit card frauds and fake reviews courses, andHR-related topics.
are two of the most significant issues that
Creating Smart Content
E-Commerce companies deal with. By
Digitization of content like video
considering the usage patterns, AI can help
lectures, conferences, and textbook guides
reduce the possibility of credit card fraud
can bemadeusingArtificialIntelligence.We
takingplace.Manycustomersprefertobuya
can apply different interfaceslikeanimations
productorservicebasedoncustomerreviews.
and learning content through customization
AI can help identify and handle fake reviews.
for students from different grades.
18
Artificial Intelligence helps create a Spam Filters
rich learning experience by generating and The email that we use in our
providing audio and video summaries and day-to-day lives has AI that filters out spam
integral lesson plans. emailssendingthemtospamortrashfolders,
Voice Assistants letting us see the filtered content only. The
popular email provider, Gmail, has managed
Without even the direct involvement
toreachafiltrationcapacityofapproximately
of the lecturer or the teacher, a student can
99.9%.
access extra learning material or assistance
through Voice Assistants. Through this, Facial Recognition
printing costs of temporary handbooks and Ourfavouritedeviceslikeourphones,
also provide answers to very common laptops, and PCs use facial recognition
questions easily. techniques by using face filters todetectand
Personalised Learning identify in order to provide secure access.
Apart frompersonalusage,facialrecognition
Using top AI technologies,
is a widely used Artificial Intelligence
hyper-personalization techniques canbeused
applicationeveninhighsecurity-relatedareas
to monitor students’ data thoroughly, and
in several industries.
habits, lesson plans, reminders,studyguides,
flashnotes,frequencyorrevision,etc.,canbe Recommendation System
easily generated. Various platforms that we use in our
daily lives like e-commerce, entertainment
4.3. Applications of Artificial websites, social media, video sharing
Intelligence in Lifestyle platforms, like youtube, etc., all use the
recommendation system to get user data and
Artificial Intelligence has a lot of providecustomisedrecommendationstousers
influence on our lifestyle. Let us discuss a toincreaseengagement.Thisisaverywidely
few of them. used Artificial Intelligence application in
almost all industries.
Autonomous Vehicles
4.4 Applications of Artificial in
Automobilemanufacturingcompanies
like Toyota, Audi, Volvo, and Tesla use Intelligence in Navigation
machine learning to train computers to think Based on research from MIT, GPS
and evolve like humans when it comes to technology can provide users with accurate,
driving in any environment and object timely, and detailed information to improve
detection to avoid accidents. safety. Thetechnologyusesacombinationof
19
Convolutional Neural Networks and Graph applications, especially in complex
Neural Networks, which makes lives easier manufacturing industries such as aerospace.
for users by automatically detecting the Withthehelpofadvancedvisionsystems,AI
number of lanes and road types behind can enable real-time course correction and
obstructions on the roads. AI isheavilyused can be used to help a robot automatically
by Uber and many logistics companies to learn the best paths for certain processes
improve operational efficiency, analyse road while in operation.
traffic, and optimise routes.
Packaging: AI is used in the
packaging industry to improve efficiency,
4.5 Applications of Artificial accuracy and cost-effectiveness. By
Intelligence in Robotics continuously refining and saving certain
motions made by robotic systems, AI helps
In the world of robotics, AI has make installing and moving robotic
proven to be a valuable asset in a variety of equipment easier for everyone.
applications. From customer service to
manufacturing, AI has made its mark and Imaging: Across many industries —
continues to revolutionise the way we think including assembly and logistics — accurate
about and interact with robots. Let’s take a imaging is crucial.WiththeassistanceofAI,
closerlookatsomeofthekeyareaswhereAI robotscanachieveenhancedvisualacuityand
is being used alongside robotics today. image recognition competencies, enabling
greater accuracy in even the smallest of
Customer Service: AI-powered details.
chatbots are becoming increasingly common
in customer service applications. These Machine Learning: Machine learning
automated service agents can handle simple, is a powerful tool for robots. By exploring
repetitive requests without the need for their surroundings, robots can learn more
human involvement. The morethesesystems about their environment, find ways around
interact with humans, the more they learn. obstacles and solve problems to complete
And as AI systems become more tasksmoreefficiently.Fromhomerobotslike
sophisticated, we can expecttoseemoreand vacuum cleaners to manufacturing robots in
morerobotsbeingusedincustomerservicein factories, machine learning is helping robots
both online and brick-and-mortar become more intelligent and adaptable in
environments. their work.
Assembly: AI has proven to be an These are just a few of the many
invaluable tool in robotic assembly applications ofAIinroboticstoday.Asthese
20
technologies continue to expandandgrowin an AI-powered system that tracks and
sophistication, it is likely that we will see analyzes employee performance data
evenmoreinnovativeapplicationsinthenear throughout the year. This system considers
future. factors like completed tasks, project
involvement and peer feedback.
4.6 Applications ofArtificial When review time comes, Alexandra
Intelligence in Human Resource receives a comprehensive report highlighting
each team member’s strengths, areas for
4.6.1. Recruiting improvement, and progress over time. This
takestheguessworkoutofthereviewprocess
AIishelpingcompaniesfindtheright
for her, makingitmoreobjectiveandfairfor
talent by analyzing resumes and matching
her team.
them with job descriptions. Workable is a
good example of this – it not only parses 4.6.3Employee
resumes for you, it’ll also use the resulting onboarding/offboarding processes
data topulltogetheralistofidealcandidates New hires often feel overwhelmed
scraped from LinkedIn and other online withpaperworkandprocedures.Nowimagine
resources,andevenwriteanemailforyouto an AI-driven onboarding platform. When a
those candidates to lure them to your open new employee, let’s say, Maria, joins the
positions. company, sheisguidedbythisplatform.The
The evaluation process is also platform then walks Maria through the
increasingly AI-driven, from video interview necessary forms, introduces her to company
insights to personality assessments and a lot policies, and even sets up her initial training
more. In doing so, you’re not only saving sessions.
countless hours of work but you’re also As a result, Maria feels less stressed
ensuring that no potential candidate is and more welcomed. Similarly, when an
overlooked due to human error or fatigue. employee leaves the company, the AI can
You can even go to ChatGPT itself to help manage exit surveys, return of company
you out, as Nikitas Marinos describes in his propertyandotheroffboardingtasks,ensuring
tutorial on how to use the tool to build an a smooth transition for both parties. It’s a
ideal candidate profile. win-win.
4.6.2 Performance reviews 4.7 Applications of Artificial
Consider the case of Alexandra, a Intelligence in Healthcare
department head who needs to conduct
Artificial Intelligence finds diverse
performance reviews for her team. She uses
applications in the healthcare sector. AI
21
applications are used in healthcare to build 4.7.3 Treatment Design
sophisticated machines that can detect Artificial intelligence systems have
diseases and identify cancer cells. Artificial been created to analyze data – notes and
Intelligence can help analyse chronic reportsfromapatient’sfile,externalresearch,
conditions withlabandothermedicaldatato and clinical expertise – to help select the
ensure early diagnosis. AI uses the correct, individually customized treatment
combination of historical data and medical path.
intelligence for the discovery of new drugs.
4.7.4. Digital Consultation
AI robots can perform surgeries that
Apps like Babylon in the UK use AI
even a skilledsurgeoncan’tperform,recently
to give medical consultation based on
an AI surgery robot performed asurgeryon
personal medical history and common
noncurableTumoursuccessfullythatmostof
medical knowledge. Users report their
the experienced and skilled surgeons would
symptoms into the app, which uses speech
fear to perform that surgery
recognition to compare against a databaseof
4.7.1. Managing Medical Records illnesses.Babylonthenoffersarecommended
and Other Data action, takingintoaccounttheuser’smedical
Since the first step in health care is history.
compiling and analyzing information (like 4.7.5 Virtual Nurses
medical records and other past history), data
The startup Sense.ly has developed
management is the most widely used
Molly, a digitalnursetohelppeoplemonitor
application of artificial intelligence and
a patient's condition and follow up with
digital automation. Robots collect, store,
treatments, between doctor visits. The
re-format, and trace data to provide faster,
program uses machine learning to support
more consistent access.
patients, specialising in chronic illnesses.
4.7.2. Doing Repetitive Jobs
In 2016, Boston Children’s Hospital
Analysingtests,X-Rays,CTscans,dataentry, developed an app for Amazon Alexa that
and other mundane tasks can all be done gives basic healthinformationandadvicefor
faster and more accurately by robots. parentsofillchildren.Theappanswersasked
Cardiology and radiology aretwodisciplines questions about medications and whether
where the amount of data to analyse can be symptoms require a doctor visit
overwhelming and time consuming.
Cardiologists and radiologists in the future
should only look at the most complicated
cases where human supervision is useful.
22
5. CONCLUSION
Artificial Intelligence (AI) is a Narrow AI systems are proficient in specific
transformativeandrapidlyevolvingfieldthat tasks but lack thebroadcognitiveabilitiesof
hasthepotentialtoreshapevariousaspectsof human intelligence.
our society, economy, and daily lives.t: TechnologicalChallenges:Developing
Impactful Applications: AI has AI that can operate safely and reliably in
demonstrated remarkable capabilities across dynamic real-world scenarios is a complex
numerous domains, including natural challenge. Ensuring robustness,
language processing, image recognition, interpretability, and security in AI systems
medical diagnosis, autonomous vehicles, and remains a subject of ongoing research.
more.Theseapplicationshavethepotentialto Continuous Learning: AI is a fieldof
enhance efficiency, accuracy, and continuous learning and adaptation.
convenience in various industries. Researchers and developerscontinuallywork
Ethical Considerations: The to improve algorithms, techniques, and
development and deployment of AI models, pushing the boundaries of what AI
technologies raise important ethical can achieve.
considerations.Theseincludeissuesofbiasin Interdisciplinary Nature: AI is not
algorithms, data privacy, job displacement limitedtocomputersciencealone;itinvolves
duetoautomation,andthepotentialforAIto collaborationwithexpertsfromvariousfields
be used in malicious or harmful ways. suchaspsychology,ethics,neuroscience,and
Ensuring responsible AI development and more. This interdisciplinary approach is
deployment has become a critical focus. crucial for addressing the multifaceted
Human-AI Collaboration: The most challenges of AI.
promising approach to AI involves Job Transformation: AI's impact on
collaboration betweenhumansandmachines. the job market is a subject of debate. While
AIsystemscanexcelindataanalysis,pattern AI can automate routine andrepetitivetasks,
recognition, and other tasks, while humans italsohasthepotentialtocreatenewjobroles
contribute with creativity, empathy, and andopportunities,particularlyinareasrelated
ethical decision-making. to AI development, maintenance, and
Limitations: Despite significant oversight.
progress,AIstillfaceslimitations.CurrentAI Regulation and Policy: The rapid
systems often lack a deep understanding of advancement ofAIhasprompteddiscussions
context, common sense reasoning, and the abouttheneedforappropriateregulationsand
ability to handle complex, open-ended tasks. policies to ensure the responsible and fair
23
deploymentofAItechnologies.Governments, replace a human,i.e, it's hard to create a
industries, and researchers are working to machine with an intelligence as capable as
establish guidelines and frameworks for AI human brain,as of now no science theory
development and use. could explain ahowcomplexneuralnetwork
In conclusion, AI is a dynamic and of ahuman mindworksandthoughtmaking
transformative fieldthathasmadesignificant process works.as of current AI technologies
strides but still faces challenges related to these can be helpful to automate tasks with
ethics, limitations, and responsible less human interventionAI uses prediction
development. As AI continues to evolve, it algorithms,classification algorithms to train.
will be crucial to strike a balance between In order to produce efficient results these
technological advancement and ethical algorithms require enormous amounts of
considerations, ensuring that AI benefits data,these algorithms are based on
societyasawholewhileminimisingpotential statistics.this algorithms won't promise
harms. accuracyofresultproduced,thismachinescan
perform well on training data,but when new
Artificial Intelligence is a
scenarios occurred this machines would
multidisciplinary field where a machine
suffer,as time goes this models would get
possesses intelligence.As of 2023,researches
more data and this models become efficient.
and technologies advanced in AI. Many
IndustriesareoptingforArtificialIntelligence
forautomating dailytasks.AIisgettingused
for businesspredictionsandrecommendation
systems and many other tasks.
A computer does not possess any
Intelligence deep down it works on bit
manipulations,its gets operated on specified
instructions.no matter how many
advancements made to current AI,it can’t
24
6. REFERENCES
Wikipedia and various aspects of computer science.
Whetheyou're a beginnerseekingtolearnthe
Wikipedia is a free-content online
basics or an experienced coder looking for
encyclopaedia written and maintained by a
advanced insights, Geeks for Geeks offers a
community ofvolunteers,collectivelyknown
wealth of information to help you enhance
as Wikipedians, through open collaboration
your technical skills and knowledge.
and using a wiki-based editingsystemcalled
MediaWiki. Wikipedia is the largest and
most-read reference work in history,[3][4] and
has consistently been one of the 10 most
popular websites.[5] Created by JimmyWales
and Larry Sanger on January 15, 2001, it is
hosted by the Wikimedia Foundation, an
American non-profit organisation.
A First Course on Artificial
Intelligence:
ThisisabookwrittenbyProf.Deepak
Khemani professor at department of
computer science and engineering,IIT
Madras,this documentation is a reflection of
teachingsofsirDeepakKhemaniinthebook
A First Course on Artificial Intelligence and
from the course of Artificial Intelligence :
SearchMethodsForProblemSolvingoffered
at swayam NPTEL(The NationalProgramme
on Technology Enhanced Learning)
Geeks for Geeks
GeeksforGeeksisanonlineplatform
that provides a vast array of easily
understandable and comprehensive computer
science tutorials, coding examples, and
articles. It's a go-to resource for individuals,
often referred to as "geeks," who are
passionate about programming, technology,
25
26
You might also like
- Hands-On Machine Learning With Scikit-Learn, Keras, and TensorFlow 3rd Edition TEXTBOOKDocument14 pagesHands-On Machine Learning With Scikit-Learn, Keras, and TensorFlow 3rd Edition TEXTBOOKrebic704740% (1)
- 03 Deep Learning OverviewDocument80 pages03 Deep Learning Overviewahmed01067393475No ratings yet
- Final Seminar ReportDocument55 pagesFinal Seminar ReportMR MONK100% (2)
- Introduction To Neural Networks by Ingrid RusselDocument16 pagesIntroduction To Neural Networks by Ingrid RusselraidenbrNo ratings yet
- Kiran seminar reportDocument74 pagesKiran seminar reportrasiphalalushoaib2No ratings yet
- Seminar ReportDocument35 pagesSeminar Reportsammuel johnNo ratings yet
- Info 23Document44 pagesInfo 23Hlory HloryNo ratings yet
- Ppt on aiDocument24 pagesPpt on aichmaheshkumar98No ratings yet
- Final AI ReportDocument37 pagesFinal AI Reportnithinmadhukunte0No ratings yet
- Google Aiml1Document29 pagesGoogle Aiml1Vethan VKNo ratings yet
- ReportDocument34 pagesReportVaishnaviNo ratings yet
- Google Aiml2Document33 pagesGoogle Aiml2Vethan VKNo ratings yet
- SmarrtDocument47 pagesSmarrtsharathchandrakaNo ratings yet
- Ai Internship Report-9.PDFDocument33 pagesAi Internship Report-9.PDFsarathreddiworkNo ratings yet
- Smart Lock 51Document25 pagesSmart Lock 51pk5178793No ratings yet
- Technical SeminarDocument31 pagesTechnical Seminargvishnuram013No ratings yet
- Gen AiDocument70 pagesGen Aiabineldho9207No ratings yet
- Rahul IniDocument5 pagesRahul InikrshxcxNo ratings yet
- Internship Repoort 2Document35 pagesInternship Repoort 2amarnathNo ratings yet
- Nisha ReportDocument50 pagesNisha Reportnishamdb05No ratings yet
- AI First PagesDocument5 pagesAI First PagesNiranjan NiruNo ratings yet
- Ashwithtredy DocumentDocument39 pagesAshwithtredy Documentvijayadubbaka25No ratings yet
- 1822pt2 B.tech It Batchno 317Document32 pages1822pt2 B.tech It Batchno 317kedop48916No ratings yet
- Seminar Report - MergedDocument27 pagesSeminar Report - Mergednikhilreddyg1433No ratings yet
- Intershipdocument 18881A12A5Document32 pagesIntershipdocument 18881A12A5Shireesha ShiriNo ratings yet
- LASTDocument20 pagesLASTx majorNo ratings yet
- Project ReportDocument40 pagesProject ReportgauravNo ratings yet
- 213T1A0416 Aicte Internship Document2Document27 pages213T1A0416 Aicte Internship Document2Harika KNo ratings yet
- Ram ReportDocument35 pagesRam ReportKonari YuvarajuNo ratings yet
- Internship Repoort 3Document34 pagesInternship Repoort 3amarnathNo ratings yet
- Anurag Group of Institutions: Digital Twin TechnologyDocument21 pagesAnurag Group of Institutions: Digital Twin TechnologyyanonNo ratings yet
- AI INTERNSHIP REPORT-9Document33 pagesAI INTERNSHIP REPORT-9sarathreddiworkNo ratings yet
- Tanmay Jain Internship reportDocument28 pagesTanmay Jain Internship reportTanish AgrawalNo ratings yet
- Anshay 22 NDocument17 pagesAnshay 22 NVijay GuptaNo ratings yet
- Madhavan M Ts ReportDocument32 pagesMadhavan M Ts Reportkrk310119No ratings yet
- Internship - Report (Vinay Patil L V)Document34 pagesInternship - Report (Vinay Patil L V)Ganesh H SNo ratings yet
- Internship Papers PreviousDocument52 pagesInternship Papers PreviousneelanjanakandelliNo ratings yet
- SeminarbionafideDocument29 pagesSeminarbionafidedadsgift0077No ratings yet
- Seminar Report 032Document32 pagesSeminar Report 032046 KISHORENo ratings yet
- Updated Internship Project ReportDocument20 pagesUpdated Internship Project Reportreddyjanakiram4No ratings yet
- Report anshDocument25 pagesReport anshjuieedebnathNo ratings yet
- ANANDHARAJ FRONT PAGE-merged (2)Document88 pagesANANDHARAJ FRONT PAGE-merged (2)saravananNo ratings yet
- Minor ProjectDocument25 pagesMinor Projectdwivediankit817No ratings yet
- KavyaDocument38 pagesKavyaKavyaNo ratings yet
- Final Doc1Document62 pagesFinal Doc1Anusha KandulaNo ratings yet
- A Dissertation Submitted in Partial Fulfillment of The Requirement For The Award of The Degree of Bachelor of Technology inDocument5 pagesA Dissertation Submitted in Partial Fulfillment of The Requirement For The Award of The Degree of Bachelor of Technology inSrikanth PrasadNo ratings yet
- Technical Seminar Sample ReportDocument34 pagesTechnical Seminar Sample ReportAkshayNo ratings yet
- SHabareshTS REPORT 38Document34 pagesSHabareshTS REPORT 38akhilesh.p.lfjcNo ratings yet
- Aditya Intern ReportDocument52 pagesAditya Intern ReportAdithya KannaNo ratings yet
- Seminar Report SAG 1 2 2Document26 pagesSeminar Report SAG 1 2 2soham220033No ratings yet
- Final Seminar ReportDocument29 pagesFinal Seminar Reportjaanramrohit1122No ratings yet
- Project ReportDocument45 pagesProject ReportAbhishek Gupta81% (27)
- Starting Template Tech SeminarDocument9 pagesStarting Template Tech SeminarSanthosh ReddyNo ratings yet
- reportDocument40 pagesreportShanthireddy MatamNo ratings yet
- P52 ReportDocument53 pagesP52 ReportTusheet Pal SinghNo ratings yet
- 1822 B.E Cse Batchno 328Document60 pages1822 B.E Cse Batchno 328Gaurav AjmeraNo ratings yet
- IndustrialTraining ReportDocument26 pagesIndustrialTraining ReportekamphoenixNo ratings yet
- Final Mini ProjectDocument50 pagesFinal Mini ProjectHarika RavitlaNo ratings yet
- itDocument45 pagesitHarmeet kaurNo ratings yet
- Final FinaldocDocument52 pagesFinal FinaldocAnusha KandulaNo ratings yet
- Smart Mirror _final_doc (4)Document46 pagesSmart Mirror _final_doc (4)Shanthireddy MatamNo ratings yet
- Introduction To Augmented Reality Hardware: Augmented Reality Will Change The Way We Live Now: 1, #1From EverandIntroduction To Augmented Reality Hardware: Augmented Reality Will Change The Way We Live Now: 1, #1No ratings yet
- Deep LearningDocument3 pagesDeep Learningdhruva badeNo ratings yet
- MAT6007 Session4 MP Neuron PerceptronsDocument15 pagesMAT6007 Session4 MP Neuron Perceptronsmanojnaidu yandrapuNo ratings yet
- Artificial Intelligence Infographic - 101Document1 pageArtificial Intelligence Infographic - 101SulaimanNo ratings yet
- DEEP LEARNING - LIBRARIESDocument5 pagesDEEP LEARNING - LIBRARIESudayakovvuri1165No ratings yet
- Figure PPT ch005Document59 pagesFigure PPT ch005Creative IdeasNo ratings yet
- ML - AI PresentationDocument3 pagesML - AI PresentationbeaustromsNo ratings yet
- SRS-DNN: A Deep Neural Network With Strengthening Response SparsityDocument16 pagesSRS-DNN: A Deep Neural Network With Strengthening Response SparsityHasham AdeebNo ratings yet
- Chapter 1 AnnexeDocument17 pagesChapter 1 AnnexeWafaNo ratings yet
- Deep Learning Algorithms For CybersecurityDocument15 pagesDeep Learning Algorithms For CybersecurityTejas moteNo ratings yet
- Generative AI Tutorial AprDocument8 pagesGenerative AI Tutorial Aprsudipkumardeonair2No ratings yet
- Fundamentals of ML - Pre Quiz - Attempt ReviewDocument4 pagesFundamentals of ML - Pre Quiz - Attempt Reviewvinay MurakambattuNo ratings yet
- Chapter 18Document31 pagesChapter 18Ali KeresteciNo ratings yet
- The-Evolution-of-AIDocument8 pagesThe-Evolution-of-AILakshmi GanjiNo ratings yet
- Deep Learning ModelsDocument21 pagesDeep Learning Modelssajid8887570788No ratings yet
- Remotesensing 13 04712 v2Document51 pagesRemotesensing 13 04712 v2Raphael NoriegaNo ratings yet
- Unit 2Document10 pagesUnit 2Lakshmi Nandini MenteNo ratings yet
- Single Layer & Multilayer PerceptronDocument14 pagesSingle Layer & Multilayer Perceptrondivejdivej16No ratings yet
- Lecture 3 - Deep LearningDocument52 pagesLecture 3 - Deep Learningrubio.deputterNo ratings yet
- Gated Recurrent Unit: Master Sidsd - S2Document23 pagesGated Recurrent Unit: Master Sidsd - S2radNo ratings yet
- Convolutional Neuralnetworks: Abin - RoozgardDocument54 pagesConvolutional Neuralnetworks: Abin - RoozgardArnold SchwarzeneggerNo ratings yet
- Exam 2003Document21 pagesExam 2003kib6707No ratings yet
- Mini Project On: Gender and Age Detection Using OpencvDocument17 pagesMini Project On: Gender and Age Detection Using Opencvkambam swarna kanth reddy100% (1)
- 1.AG60202 - AI Applications in Agriculture - Rajendra Machavaram - Spring 2022-23Document26 pages1.AG60202 - AI Applications in Agriculture - Rajendra Machavaram - Spring 2022-23Ajwad JumanNo ratings yet
- Jon Krohn Metis Deep Learning 2017-05-01Document107 pagesJon Krohn Metis Deep Learning 2017-05-01Arnaldo Preso De LigaNo ratings yet
- ANN & FUZZY Techniques (MEPS303B) : Q1. Short Questions (Attempt Any 5)Document5 pagesANN & FUZZY Techniques (MEPS303B) : Q1. Short Questions (Attempt Any 5)nadeemq_0786No ratings yet
- Adaptive Bidirectional Backpropagation: Towards Biologically Plausible Error Signal Transmission in Neural NetworksDocument11 pagesAdaptive Bidirectional Backpropagation: Towards Biologically Plausible Error Signal Transmission in Neural Networksgheorghe garduNo ratings yet
- DL Mannual For ReferenceDocument58 pagesDL Mannual For ReferenceDevant PajgadeNo ratings yet
