Unit -1 No

Date 17 I 1 2 2

AI Domans

Domaios to AI eJen to main branches A

o AI
T
dhey ave Hepre sented_in the follooing Chast

AI DorainS

Data Compute Natnal iNLP)

Science vision Language PJoeSsing

Data Science 3-
D.S is are aboutaPplyin lathematical and
Stat atic pripal ndata ozinSimple uwo2d
data Sience s heStudy ot data thsdata
Lan he ot 3 types

Audio
h) Text
Lompufer Vision8- C.V_n_ Simple 1word u
identityin9
cbject or
Zhe Symbols from the giuen
image leqn the pattean_and
Predict future aut Outome
AppJitation of Conputenyisions-
face falter
 *****************************.

Date 7/ [1/22

) Uqoogle lens
(u) Rtémotive ans self driuing La )
) cogle dhanslate

Ppplhation of lomputer based on a n0 of dask

that lan_be pexfos med on input 9mage qnalize
t
and predict the autput.

Natural lanquage
Ihe abtlty_of lomputen t o understand human
danquagesas kaSpoken 0 wrten and to qiue
Outpuiby puocessing it salled Mahsal Jánquag

NLP has 2pazts

NLU
( NL
NLU Lollecting the
is Lalled intoxmantion
NLU. andunderstandhg
understandg itil-
NLb Th9nKingof
alled appropsiate_od
the uper is
Matural put
Out Put D ked
eEd
Jargvage_qeNxatioon
 No.
Date 7/ Ii 22

3 Appitato of NLP

(Speach necoginatton - s when a System2s

able to ue Qutput by understandingor intesp
USes Speah.

Ex. la- Assistant

Alexa
Lortana

Sentimenlal dnalysis
Sentimental analysisS q Pancess of detecting
negativeOr positive Sertiment in a éexé
or in gnymage..
T s used in youtohe uoiltent qaphic taltent

identifytrng Scam me.ssage. blecking cvt Social

mediabst.

)Machme tanslation Tzanslatian af a text in4

Llanguage to another_dan9uage hy Machine
V
 No.

Date /8/II /22

AI Cthics

Moral issues
y Data pivaty
AT Bros
AT Access
AI Threats
AI Ethicsare mal principle that goueana
pesson behaviourozthe Lonductof a act?uity

he ethics of AL j the banch_of Artiicial

ntellngenLe SyStem Hespansibl?ty

Moml fssues
The moxalty of developer gets transfexed nto the machie
as WhatalCozding to him Lhen is ght
andhence
would be the Selectian made hy tho.machine.

Data. pzvaty
Data psivaty Sometime also
priventyJt ts an axea Mefered
0f_datq
as inlonmatian
handling Sensetive data and Psotectianand
misuse Prevent_t fon
 No.
Date

AI Bos
Bias retersto fauawning of one side OUen
anotherno mattex hou much. one tates to be
on_bias.
Majoslyall thevoice assistane have atemale
voicen it

AI ACCeSS
Since AT1S aAec_ echonolgy hence it s
Lostlfe.. The peaple_who Canoffovd AT qse
Ach pexSons Beause of this a4P d energed
between two classesof people

AI Theats
AT Lzeates unenploymentas AT 2s Making eoples
Pfe easienheneádhe dabouODS obsade
being e placedhy machines.
 luit-2 No. *****

Date
AI PROJECT (YLLE

A.L project ycle s a Step by Step prtess that. aa

USe follaws in Oxder butld an_ALprejet Jhe_
to
AL Preject Lycle Hers.a hat
Sunble trmeawork that
Lan
Lan guide s h g h t direction.NsLycle
nencdlvdes 5 Stages

Slagei Stage 2 Slge 3> lage 4 Gtape5_

Probem Data Data Modellitg Eválstion
Scapang Lqiisition exploaaticn

Stage 18 Prablem Sup

Stanting
Starting uth pmblem Stoptn9 yov detfne_the _bjecties-
of jou Psefect_by identiryiný the ssues yo hop
to address.

Gtage - ato acquisttien

you Need to acquise data uJhich_will betome the
the hase
kase
of youz paoject AS t will help you in_understand?g-
what the parameters ae zeated
t your obfertiUED
Stoge 38- Vata explooation
CriLe the data you Collect anld be
in_darge
quantiteswe w NepNeSent the data inYazias
tn Ya
fom Lke data base, flow chasts ett.
In this Stage we f?ltexout the
gxaphs
hbicate_and _u/10
 No.
Date

data.

Stoge 4 - Modelltng

In this Stage we detesmne the Mocdel uwe axe

uSing or au 2aofect They_are_mainy af
twd types

1Leaxning based madel

2on-lasniug based niadel_ (Kule- Based Model)
h9s also ncludesthe hree_danains_
his A.z

Stoge 5 - Ealuatlon
In thisStage we test_au model by_wsingarious
techbiques Cnltne aS wel_as n HEne
Mode.
E v a luati m sbase d on 4paza mEteAs
DTr Positive
2) Tue Negative
3) False_fositive
4) False Negative
 No.
Date212 22

need to
AI poject tycle Jnepresent five stoges uihich need to be
he
follouoed n_ovder to altopmpl'sh are paaieLt \he Stoges
Stages
are ShouD belouw

Stage 1 Stage Stage 3 Stagey Staqe 5

podern Data Ddt d YodelinyEvaltion
Scoping Acqusiton Exdoxatipn

hablemSco2ing

Det ning_what o do do sohe a problem

Pooblem Scoping 2s the vexy f?irst Stage ans_ AI
ofect in this Stage wefinalts e the atm?ot
OvrQojet.As an individua yau Can off foz qny
for any
Aojeit but as
aCopany Or form you need
fo_tollow STnLsustalabl develapmentbjoal) hy
Dntted lation These goals aze of 17 dffesent
types Some ot_them ane

1No oveat y
2 Zexo_Hunges
3 bto0d Halth and wel-being
4Bpalty Education
5Affordable and Clean nesgy
6 Industay Tnlouation and nfrastrture
7Clinate ddion
 NO.
Date 24/ 12122
8 Redeed Tneqviality.

Ws Psoble1n Canvas
The his hetp in dentitying the key
which 2s Helated to Ou rahlem.
_element

Ws Problem Canvas
.

Who-
In this Section_ue analyse the 2EOPE uho
ane qeting_affected divecty or ndiuettyy hy_the
oblemThese geople_at dlso kncan_as
Sta Kehldeys"

What
Undern thisSection_we discaibe the_natue ai the pablem
t_ lan be tor a Shoat team_or tor lcng tezm_as w2l,

Wheve
Unceyou knouw who is aisociated aith_the Poabtem_and
what the problem_agtually s_you need to focOS
On the ontext lotatioh Istuation 4 dhe poblem.
 No. - * * * * * * - * * * * * * * * * * * * * * * * * * * * * * * * *
**

Date 24/12 122

WhyIn this Section we_descaibe_the benefits_of

CU pxoject gele and how ate_aject dfett the
Stuntiort In o sitive manney

Wata Nayosacin
Collection of data Compltlation

The Second stage_ot AI pooject cycle--s

Data
Atqufsition as the tem Suggest n_th°s Stage
we Collect data foo ou and his data
will beome the s e of
p0ject
Ou paoj.ect.
We ocuide data as input to oo System.
These taininq data ha s to be Jela
authentic ble_and

Fo toaig data coe_an visit these wels?te

data gou.in
IyTndid q0v. in
S
 No.
Date 257/2 122

DATA EXRORAT ION/DATA VISUA

A
Sapl2tPed
Raw data Pipe ne
L.SV Ples Understanding
xISx f?les haessS clata.
txt fles

he ipe ne pocess in a SetoAction_which

'dataMo Simlied undenstundibyakezh
Changes daw
for's

Data Exploration s_Zhe qaaphtlal_veppesentatioN_of

nJoamation Rata visuallzatton ftol Prouidesa
wal to see_andundezstand dhe tsends_and
patten between_difexent data bets.

ADYANTADES OF DATA Exploaatio

T1s Used fos amly sing mastueamaunt ot
data

UDTt s the best techiiquve to doaw dyends_qnd

patesn anong data set
U) On the bases ot data ui sua liaation_an chantB
gsaphs labes, Dashboaad
 No.
Date 25 / 2 / 2 2

Soge DELLTN
Rule basd hen Role Ba sed
(leasning Based)

Atogeu_gk P3ofect cycle that is Modelling_divstbes

the type"ofmodel we ae_usind for OUr
psoiect
hey
hey basirally_tun types_of todel
) Role Base
10 Nbn Rule_Based Ldeasnleg Bas4)

Rule Based odel

As dho name suggest he_deialepen2
deinesdhe wlearinstxucticns foy dho lhofect
and ted if do dhe_machine along wh he
aaining_dada
Jhe_only drauback f this 4ppraach he
becomesfaticsHente Cne_he
leaning
machthe sained Lwith thls_data_anncz#
change tht_Output.
 No.
Date 25 12 22

Example 1 Cnline_hess game

2 -10aJHex

Non Role basecl Madels: n leazning ased Model

cthe
Machipe dhaned th_dato dbng wth
Scnminstavctionsand then abl_d0 desaan
Cortut aaosding_ Ahe draning_dat2

Leanngased

Aupenisèd UnSupen Reinloxement

Mcdel Uised Modee
ipdel
T oIP feed back
dabelled
dau

supenised dah
Supebuised leasnings dhe dype er
tegattng in_ which machines
well Jabelled clata he are trained usiog
Some mput Ps dabelled clato_meat's
Output
abeady etaged uwh laxrect
loxrect
 No.
Date
251 /21 22
Soer uised macel has To type.

1ClassiH iead pn
2 Keqession

Uhisupenvisecl data
UnSepeviseel leaning_Mldel cwoJks on undaber
_data sef _Dich Means_horiachineis Jed
2th Andamnut his_moAdel has the lapahiliny
Intespa etny the data _and make ifs own
algonithm

DiSupeauised data haslao dy/es2

1 Clostesin
2 DimenSionalety Reduction.

e inlor ceent
Lome_dimes a xeqvied piodel alue
bt dhe eutput_is 2ot Known buf fhe Sycten
ProuideA he Joedbrck Called rein,Jaz eiement
Jearnin9
ple - weather /osCasting
 No.
Dat
27 (2 22
EVALUTION
Evalutions the last Stage ol A.T project cycle
in_this Stage the testingdata of the mathine is_
ealuted ch che Jallasng aameteas.

AccuAccusac
Li) Recall
V) F1 Score Uit-6
NLPEMATURAT LANKle PROU
MP Atandy or t is a dechncloq4fhat Tt used hy
Machne t ondeastanding anakse ancd infeapret husman
danguag esit an be nuaice lomandor in_Jextual.
tooth.

Advantages LP
MLP hels uses to ask
question_and qef a_direct
TLYOSe whin a Second._
LP cHess exect ansiersfo dho
question wiho
an unwalted nbonartion.

u)Tt B ery dme H'ejenta

 No.

Date
2712 22

Disadvartages of NLP.

1NLP maylot show Contexts

2 NLP sppedectable
3NLPis vnable to_adept new domain

NLP

NLU NLGT
These are_tuo lompoIept ot NLP
1Aatusal_clargunge understnding.
2 Nadtusal datouge gentExatioh.

NLU heps he Machne_ Lndensfand agnd

nd

analyse human clan9uages_hy extacting_ho_ain

daka oom the Contént.
~ihc
NLG
NLG7 acts as a translates cthat lonvezts
lomputerised data nto humandpec?fic
dangcages
 No.

Date

APPLTIATIN OF MP

1Autemati mmdzization
2 Sentimental Analysfs
3 Text Classiffeaton
4VoiceaSSistale

Autoontic Sunuarizat0n2
his s g PE n vhith hupe_aount_OÉ_
text data got Aumarised in jëw pqes._

Sentinentl Analysis 2
The goalo sentimental Analys?s isso
Rdentiyentiment amaag Senesal 2ost.

lext lassALation g
Tt s a 2iie t Cateagising ext MtO
ngan?sedgovps 03 pre dene text
Vcfte asskstante
T s a technaloqy Usng_Which
Lommvntlade withmachnes. we Can
 No
Date

Chat bot 8

Chat hatsa A.T fechni9ue hat vse NLP do

ondenstad Shat a human uants andguides
dhem do he desíse Cutpot

chatbet _mist hque folowing eatuzes

T sesoleasation_data etocndesstand t
dye at questia).

(1 Aer anadlysinghat data it gajerates a nespcnse

(uD Chatbot USe machine deasing ado_Jegzn dhe
(onHext.

Hee he hepe ae duo types of Chatbots

1 ipt box
2 Smasdbot
Sasiptbot Smartbot

sipt bet is based on Sriastbot_uses_Ma chine

Sdtpted Pgzam eaanng on_dhe blggen_datal
Sets
Scajptbot is used s w Snilastbot is used op A1
messa ging platos enable websttes

taited fostionaltty y wjde funtionality

 No.

Date

Text nomaksation
Jext eonmalisation ?s4 pIefpaOeSsillg Step aimy
af amed inpnave uhe qualilyofitpt and
Making it suitable for machine_do PIo1205.

5 Steps

Takenizatiop Coamon
oopus Sentene CaSe
Stemmany
begmeshetion

LoppouS Colectiwen_ot_wsitten_text is known as_(oapas

T may be of mulEple pages this i_dtho intormtion
USey ouides to he dachine

usothente Bequentation 2 As_ho name su4gest loxpos

s dyided Mto seveal meningful sentantes
n)Tokenization n dhes Step_dho settences_i
foahen diuded into _wwends.
( omnon Case_A Ahese_onds aze in_aauncbm_cas
hich heeds do be Changed
nto louwen_lases
 No
*****

Dal: 24 12i 22

Semtog 1emmattzation2
Th thisStep he JEman9 words au zeduseto
200t wosds Jn dhis Prdeks_h al?Kes ot ha
woxds Q HeMOVe.

NOTE 9 Th Stemaing dho oord changes to

oot 00d but i t dae.snot mean_dhi_L0od will_
be meantngfal.

emmati2ation stemmng and lenmatizatian_are

tho Sae Ioless- he otly diffexene is_i
demmatization he wod hch _is foJ is
meaningfl.

BAGs a oRDS
Bags _ot wods is_aWLP alqosthm shich helps in
ektracting ttdtwes _Cut ot dha text uhih coill be
helPrul in_machne leaaning algosithn fox this
unpose wve need to lonert tekt_data înto t s
lemericfoom.

7
Bow he 5
algo itn She
and=3
Noama lised
datq
 No.
**

Date //

sAs you LanSee n the mage ofwr

Shown_a a_ lfst of wwords
the Aumben (oresgondio
ppeaxing in the lopus_and
to it show S tho AUmben ot time the uoxd
ha
been sepeahed.
Now we lan Say that buqs ot uwords _alhoathm
gues two_thinps.

ValbwlasyTmeans_ the_onigue wonds qpreSent_in_

the orpåás

Fxequeny T mea0s hou many_Limes the uwoxds s

zepeated
talloingaw_che Ateps to immxtant ba9 of
Wods a lgoaithrn

L Noamallsation
ly (1eatng_Ditkionany
u) (3eatikg latowed YectozS.

Example_Of(Bow2
Step 1 Mbsmalisation

Dor17 he lat Sat

Doc2 Tha Cat Sat n hat
Dot 3- Tho cat wih hat
 No
Date

Lomplon lase-2
Dot1 dhe at Sat
Doc 2-7 the lat sat n hat
Doc3-7 the lat wih hat

Dacwmeyt lectos table

(Dotument frequenty )

the Cat Sat n hat wth

Oon 1 1 1
Doc-2 1
DoL 3- 1 1

finally_dhis 99ue us dhe_docutenf ector table

for ouz lorpos.
TF-TOF Term tequencyInease docoment
epuenty
Team
Team freqoenty Jnvezse alaoment trequenty is dhe
method using_61h ie eualoate dha mort ante of
wd_ndocumett The TE-DE Scasiig ctha ualle
intaease in propbhion do ha namben ot mes
a u0ICd apea7 in dhe dotonené

NOTE Remoal of Atop uioads.

 No.
********~ .

Date

Stop wonds ae neqligible_w0ds awhich donot he haue

any meaning ofits ouin dike qy ticles, Conjurtionete, etc.
hese cwokds_ae Henve by his algoaihm oz
0 Pre -prbtessing Btaye.
stoy ooOd

OCcunence
hregenty
Razeoods
Value
TE-IDE

Jesm trequenty 8-
Tem heqUEnty is the requenty ot a _uwork n_onN dltkner-
Jerm tregoency lan_easily bejaond fiem dho dument_-
Vectox dable.

Dowment trequenty_ 2- Wolament freqvency s her-

the_Au
of document in_which the words ourente ixesetive
p how Malty aimes it has otLUed in the docwment

fomula of TF-IDE
TF-IDF TF x lo LIDE) (tesn)
whese,
TE TesM thequeNy ot
a woxd
OFSommation ot all TF in tha
dotumentt
 No.

Daie

IDf-Nambex cf locument_
VoCoament frequency
Exanple of TE-JDF

Atep 1-Jevm tsequey

the lat Sat hat_h
Do 1- 1 1
Dbt21 1 1 O2
Dot2 1
Step 9 ktoment Jaequeny

-the iabs.at n hat wt

3 3 2 121

step 3 Tnverse docoment Jrequeney

the Cat Sdt i hat wh

Atep 4 T-IDE-TF lteam* log

LIDE tesm))=1xlog l)

Applktation of T-IP
DoLunent lassifitation 2 Tt helps in Llasilgin9 he