MI W

Page Nn
YOUVA
Date

Iatmducion to A Fourdatianal Concepta

DArticial Toteltgence
Man-made
Abhy to_percoine or inter
io ormatn and to metais t as
Koouladge to be agplied tocoads
adapire bebaviouts cothi
enironmeot or Context

> Matheroatical Lagical

’ Liogistic ate igencz Reascoing
’ Spatial ViAual ntelligeote
’ kineasthebie potelligence
’ Musical otelligeze
’ Pabrasaal 2obeligeoce.
’ Bisbeotial Poteligeoce
’ Natusalist Pote ltigeo
’ Poteypessnal Thteltigenca.
s) bat is satithcial Tobeligeoce2
> aility atunan- made tstofate
Cnachine oY
ke iotelligzica.
L4) How do machines become Aatici ally
2oteligeot 2
&CateaorieA.
Machine Lening
’DoroQins
Dala Sinca
 a Cooputeu Vision
B. NIP Nalunal Language
Larguage Inio,
AD arund us
5 Apications
2 Cootet Crcation
Chatbot% 4 Virtoal asistan t
3
AAutoDODDLs Vehicles
bal pusi
5 GFS- GloMaike toning systrs
6 Sales 4¢ tr
Loguia and Mply Chais
q. Agrtcltuxe
lo Röbobs

) Ar, ML 4 DL

ML yMachine Learing

)A Phics
’ Bias and Riunes
’ Accouotasi thy
’ Safeby
Humbn- A? ntesaction
’ Tust, Praracy and Cootrol
’Cybe.Becusity Maliciow Use
-> Atoatton land eopact orer jobs
’Danooay, ctil ngha, Robot nght, e
(5 Slages
BProblero Sopiog
Data Faploraton
Modelling
5) Evalualbo
DProblero 6coping
’ 4WA Problero Canvas
Who?
What?
Wheu?
Why?
Pooblero Statement toroplatebunnoaLse
al the key points ioto a sagle farplat
AWs.

2)Data Acgusiton gathering cata tos the

Apeuitton
pxajet ’to tain troata
the data
’o get testing data Cprudic ton
data set.
bye of cdala cant to
Data teatua otlect:
Suaveys
websies.

’ Ap2- tqplicafio aagarn Thtenyaa

 Data bxeratioo
ylelaus Data isualtsatbo
Data likllsato-epreotng data i
ghaitally

Bar yrapb

Maps > Chompeth

heat map

Netor

akgraphics hond clod.

Statttr chert
Bubble chait

AMode ling Maching ates f analye tie

data
AI Mocals
Rule Based Laazning Bate
ppnach apfroath
 Bakd
Supenised laosoing
Clasitication
k Clus teaing
a Dimanszokity Racuction.

5) Fvala lioo
Data
tes birng trained and Lindergcre
’ ecten Cy o the besting dat a
model basd on thl Pasame le
Acesay
3. Pecall
4. ELScore.
Nausal Netaes.
human bxais
’A Neuzal hetork is diided
isto muLtiplu layex and each layer
s tusthei ciidtd esto serelal bloCkA
Called Doces.

Prput layes - acuire data

- Whole
Cnot isible) pro ing
B.Outpd layo- gines user theoal outpd.
 Pape No

Date

teaturcs t Neusal Netuork.

y Mode Llud on thL humao bain and
hetvous Bystro
’ Autornatiaty extract featues
eltheut Eoite
ibut fhe programmer
Evey NN rode s esenially a
M alganithon
Luselul cohen
agedata 'solvicghet
pxoburos r ey
 vOUVA

Data ScienCe.
’ Workinq axaund numeic and
alpha..numeic data.
analysing.
analysing the dada
Hhe machine
ohch helps in
otelliqurt
to peorto tasks B, telf.
Aopicatior e Dala sâunce
|: Foraud f Rsk Detecti oo
Q. Gene tics GepDroics
3. 2ntenet Seaicb
A: Tasqe hed Advestising
5 Website Retarobncodatoos
6 Ai oate plamig
2Data Co lectioo
3-Soua Ces af data
’ ofine data Collec ion
Bersos
Pnteaecs
Obsevatbns

’ Onlre Data ColluctenGoveMent penrtals

open Sousad
Re liable hebsite
Organisattor' Open-Sous Cad
alorld
Statstfcal ebsites.
 ale

A. Types Dada
csv- Cena sepalated Valus
-y nacsheet
ySsL- StmuctusadCucy Laguage
5Data ACLes - in Tython Code
NustyNumaical python
behuecn
Pandas- Sstteau Gbxoy fa python
pacganoing anguagt
Matplctlib’ isualiza lonihnxy
Bas gsaph
Histdgiam
Scattla plot
Area plat.
Pie plot
Basic statislits with
yhon.
Mean- ho avelage Value a a
degue
Median- 5) pentantle Value ota seguence
Made - Mast drcqust vaue of the
bauente
Stardard Deviabco - Meass the spmead C
the seuance around ib aernge yalue
Variane - Average of the dgued
dilercres, fo the mean
DData Visualisatian
y While
the dataCollcirg data, i u passble that
Come wilh Borm eras.
Eoneous Data Qdays o which the data
Can be eiDneous
> Poemect Vales - Valut io th dataset
at andro plac
irioTnect.
-> Tovalfd ox Null value-
AHScrne places .
the values aet corupted
Mising Data- IoPo some datases, Bame cel
Xemain eropb
Hissing dtt
data Caet be
iotefne bad
B-Cutlies Data cohich does Dot all in
the ange a Caxtain element
 Age N

Lomputa Vsico
Asnlcalins f lonitu Vsion
facial Reuagnitioo
3 Gogles Senzch hy îmage
4 Coroptu Visioo o Retail
5 Sell axiing Caxs
6 Medical hoagiog
cooge ranslat ap

SClasittcation
’ Clascabioo + Localisa bon

Ror Malliplk objeck

>obect Detection
Sgmuntattn
clastcatn- asigoi cg an dopt inmageon

dasiicatioo 2daotiying obat ohiet s preent

- Tdaoliyi at oha lacatin.
Lcalisation
ohject datection - rding instau eal
bicyces and bilcling
 W

Page No
YOUVA
Date

ostance egmantalt - pocass o delec irg

4iving tho a Cate gory
giie each pul a'abel oD
the l bas that
Basics e roages.
piae- pictune elereot
- Sboall unit o Pofonoation
- Dqsid.
" Resolufico- No-o! Pela to ao roage
erpres mesolu tin as toidth by
height
No-st piels asas a Sicge nisbe

1280 CWidt) xl024(heijt)

Prael Value-descaibes how batght hat

pizel sardlo cObat
Colous H Should be
piul foat- bybe roage stoscd as
an 8-bt
Values roro o to QsSS.
>0 s to be tacen as no Colo O
Black
725S s tan to be ull coloui óx
Iohite.
 M

age N

Ghauscalk mages
>Mange sha des
apaunt colou. Uorh
’tno value s piraal- Daiiest posible
Shade s dack
Shace Ls cohite.
otetmediale Shaces
eaual bihbey
3

coloss
Jay scale eachpiel- Lby te baa
Sibale planla
RGB Pregea
Pedneeo and Blue.
’AU the Colouls that are
cor be nade by cOobioig ePresent
dilent iatostes
qeenad ble.

Prnage dats
cobch
s uleaot tox Bolvicq the omputatanal
tasr aelattd to Catbaio aplication

obËecta
 YOUVA

ate

Nalaal languae Brocesig LNLPD

Aspie atiurs cy NLp.
-y>Autormaticsummaiaton
Seriment AnalysA
>Text clasitica ton
’Viatual aSt ants
AChatbots
>Mbku Bot
>Clanei Bot
>Tabberacky
> Rose
Ocbatbot

2 types chatboti l. Soript bot

2.Soat bot

Scaptbot
Easy totoa
oake

iotsgrat
2. u andeasytolaua
4. No ca_iH faxocesing silla

Srast bot
esible anol fajl othel iesarces
olatabases and
2. ceork oo biager
oliuctlycoth mone data
3 aao

5
 ato

Haroan larguage le Larptilonguage

Hrnan Lagage
Coouatto thzeugb Larguage
Hanoac bxan tocps co pesing the
Saundk tbat ¿t hias

al lhe rroe

Copt Larguage
yCocpta Clodesstad the rbe
’ Every Bicq sked to toachne basto
be conexted toerobey.
and loes rot paoces
> Coruriaios oade the
boachines aie Vely basi stocple

Syctar - qzatoatical stnuatuae of a

Bestaoce
Seroactis -
Dterent Syrta, Same beraatu
2+3 3t2
TStent Berantiu, bame bgotaz
2/3 CPythor 2-) 2/3 CPython3)
 Multiple muanngs of a liord
egtace tussed.
Text Nomalisatio.
> Sereral steps to pormate the taat
to loweL Lavel.
7 Corpus - co llection of ttot docuaest3
bunch y tuxt iles en a dectn
Steps:
Sestinte egmentalion
’ oboe Corpus udded'oto sentences.
To kenisation
Each beotence tuthe oiided into
to kars
>To kars - texm wsed tor any coord oY
buro ber ot Special Chasote
Renbingstopuocrca, Special chazactzy Nurobess.
stowords - 00rds 0ccut. tugustly coHch
ot add
do oot any value

Connetng tunt to a Conan Case.

Cooveting cohola tent oto a Sicoila.
Case, plejeably louei
laue case.

Stering
ate reraal , steramoing coxd is
meaningt Or Dot:

enmatiza fior.
>attesenoYal, word gives meaning.
 M

Pago No

Date

Baa of Words
’ Step by step apuacb to ioplenert
bo cords alqoithm
Tent Normaltsatior

3- Ceat Documiot Vetos

Step ): Colecing data and pre-procesingit

Alter text noalsatioo.

Shepz: Ceate Dictiooay

Step3: Document Vector table.
 Evaluaton.
Evaluatior
->Frocex ot undestanding the
bility 4 any AT mpdel, b&d on
outputh by tectiog test dataset into the
Mcel ard congpating cito actual ans wes.
The Conyusion
Phecic
hon Reality
Matö Yes No

yes TP EP

No EN TN

Tp- Tae fositine

YRYEAhe Paecicfon Peality matches - 1ue

Fase PaAttine fale

Bedictian ¢ Reality does not rmatch-

Tuu Ne4atine Match-Tru

Preciliorn ¢ Raality coes at
Predicton u fake - Negatire
fadse dlo
Negatie Reaihy do rot match-
Fale
Pore
Ponedictiorn he- Negabie
 M

Page No
Date

) Evalua tion Me thods

Comect Praclictron x ID0.

TotalCases

tre + TN) Xloot.

TP+ TNt FPH FN)

Precisior = Tue ositine

Al pucictadl pafitines
TP
Tp+FP

Recall Tuu Posthive

Tu postie tEalse egatie
=. TP
TPtAN

Precision x Recall
PorectsiotRecall