Assignment No: 6

Name Puua anicy adhav

Neai TE-TI
Roll no 1T-302s

AimAssigoment Not Assianment on multi laue Neuta

netuonk model

hoblemStatemernt Doonoad the dataset of national

institude of diabetes and diqestie and kidney
disea&e forom below link Data set : https:/lra.
qithube2content. comfibotonler Batauetslmaute
piena -indian-diobetes. datacsv.
the dadaaet is hou total g atin bukes_ cohere +he
last atmbte is elas attnbute " having Values o
and iCi: "Posidive foe diahetes 0'Negative ")
aLoad the dataset in the poogcam Define he
ANN model coith keas Detine atleat teUO
hiddes lauexi Speciy the pelu funchian gd activa
4ioo uncionfos the hidden laues and Sigmoid fog
the ououd laue
b. Compile the model coitth necessaey çatanelezs. sek
the numbe patameteea set the numhe of epechs
0g boteh aine and ti the model
c. Evraluote the pefremance ot the model fot
di tferent valuei oE epachs and bateh stae
d Ealuade mdel pezfoemanae Wai nq different
a ctivathoo fenctians viSUaliee the mdel Uling
ANN vìsualieL
Thecu eAniial Neuna Netwank
- An antificial neunal netuok is inspined
Ahe biclcqical Nvo Sstem
The wa by (ohich bialogicat nervauA Systenm a
nain mmcesses the infomotioo in the sor
Qi
manne AN ala pmceise the intormation.
-Anl is also calledas iofonmatian poocRAS iog
panadiqm mesembles the ain lo too eiped
leoning ponoces is uded to acqÀTe the
knawledqe fmm the envmonment by the netuon
" Acquied kaoLDledqe is
le stoTed alng Inteng
sing
connecton shengths knoun a synapic ueighBs.
adit
- Biologi cal sslem leoning poceys ìnvolvea
exist
at
ments to the sunapic Connectton
hetueco tne neon,nthe Sae mannes of
is commed o t in ANN.
leaning
ANN comptadions may be cme Out tn poxalle
and Speial hordume devicei are beng desianed
nanufacBuned cuhich takeadvantage of this
and
capability
Sigmoid Neuon bina
siqmoid néoon Can he uSed fox bot
closiEicaien ond menression problens
outÍut uoiIL be a Nalue oetuueeo 0 and t
he can use t to
and oa ed on he Value we
 this claseiEicati o bsed ao the threshold
Io cale of a t dinoensi ood iaput , tHAe
aigmoid function that besd descn be melationship
betueen inp and output is qe y

CaLe of 9 -dimensonal ioput, the stqmoid

Euncion that best des con bethe tnput-Qtput
elation is qiven os
-(wi tue2+b)

In case of high dimensionalinput auth many

featuegthesiqmoid feencan is cqiven by
(14 -(wl+ b))
e

# Recifiedineaz unit(Relo)

max 0net) O e t n e < o

-ReciCied ineqenitsone of themost frequent
Lused odyatio funion in deep leatninq
Relu functoo s a fast- leanine activatiaà
func4ion thd assured the delive z of qood
pecformanceuith appopia te and Coomect retts.
Relufunctio0 peoxme Q thresho ldioperation
on ech input eenentuhere all values less
 thanZeLOaeSet tO 2e2 o

# Tanb /Bipolae function continuoesi

-net-|
e

the slgmoid tansfe function shown abare

takes the input cohich cap have ary val
betueeo pls and minu in finiy and sguas
the oLtDut into the ranqe -t to 1

keras nd ANNi
- kexas is a hiqb level neoal netuosks ApI
unten in pu thon munning on tap of
tensoflow
1 is detined to simplify the building
aining and building and developmen ote
deep leaaning mo dels
F1 moyides an easy to wse intexface forr the
bildiag vanous tpe ofneunalnetuooxk,
incuding ANNS with vezy few lines af code
Lepochi
duing traininga epoch mean that evexy
tainiag Samplein the data sek has been
Used o¯ce to updatethe wei hts.
in typical deep leanlag model, taaining te
 saely enogh fonit toleoxn Seful
pattens. therefonethe tairing pocess
0S Mepeated ovex several epochs o llaO
the netwok cto convexage to an opdimal
Soltion

#ANA Vìsuali2eE
ANN yiSualizet s oa tool used to visali2e the
aachitechue of an atietcial neumal netuook
the sBmchue
t helps in bettez cunderstanding between
compositioo and Connection diffr
layer
neuson in netwnk.
Gaphical Repes eatation: +pvide a visual
how
depicion of the netwonks layer and
neumoni qne conneted betweeo lauee
devel oper
£ase of undeystanding : helps waex and
undestand ho o data flons tomugh the netumk
Qnd how layees aone stnuchuned
potential
model debugqing Can help in dendifying
iseues in made Stouchwe dumng development
o ohen optimizing models.
- Tensoxboogd is tocl fox isualizing ANNCNN
and che neura netomk tupes
modeloptimization iS Ver Mseful thìng we
qet fom AlN isualR2e2.
MiA Lals_ane qneat fm poresenta ion and communik
For die dèppe to coplexity
Concluusbo thu& thethe AN madeland functiona
Melated to AIN raodel and
undexstand ANNVisulio
and implemen ted suCcesstul