Scribd
Upload
104 views

Terminologies of ANN

Uploaded by

rioraj0012
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
104 views

Terminologies of ANN

Uploaded by

rioraj0012
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Page No...

19

1.5.TERMINoLOGIES OF ANN.

) Ioputs : Source dota sed ito the neural nehn


a dec(5ln Y
wt the goal of makig
pretehon about the data
Tobuts to neura) network ae hyprca ly
of real vaua, each value is fed
set
loto one f he neurons in he plaen
the
2) Troming set A Set f inputs for which
koousn, usedto
Correct outputs ane
tan tthe neural netwooks.

tuworks generate theiY predictor


3) outputs : Neural ne
set of real
vaue Y
to tthe form of
value is
boo lean deasions Eoch output
huons In the
generaed by one of the
output laye.
perceptyo. The basc unit t the nural nlo
4) Neuron generates a
Accept an inpt ad
predrcton.
tach neuron acceps þant ofthe
7oput end passes it thoough tthe
activotion functor.

5)Weght pace : Each huron a nemene wergh


The weights, togeth uwth the
actvaton suncton, dene each neurons

Classic
Page No. .20

outpt eural networ ks are dratned by fne


tuning werghts, to discove the optmal set
weights that generates the raost acaurate
predichons.
orwud Rss: The Sonwd pase takes the inputs,
passes them thraugh the netuork ancd
allous ea neuon to reat to a fachen
to a fracton
fractoo af tte inpts. Neunons
generte they outputs and pas thern m

to the next layer, uhl eveually he


netwrk generates an Gutput
hEror funchon: Defnes houw far the actual output of
he Cunet model ie from he corect eutput
ulhern he trainir the mocel, the oectve
te to minimxe the e r r functon. ard biog
output as close as posbe to t e C o t
valie.
|) Backpropagaton: Ib oder to dscover the optima)
weghts Jor the heurons, backurd pass
pertomed , monng kock from the
netork's predicton to the heurons
thet qenerated that predchons. ThiS
called bck popaqation

Classic
Page No. 2.

a) Bias and anance: klhen trainng neural networ ks,


bìas and vanonee are be 2 man

measUhes used
Blas measIhes bow well he mdel

fia the traing Set. helps to predict


The knon outputs af the tranlg
exarnples.
measures how wel| e mdel
Vamance:
WoY kS wtth unknoun inpts ttat
were not avatlabe duing tralnig

paaometers'. A hyper paameter isa sehg


|(o) Hyper stuctue or operotn
affects the
netuork. I t Is-he
of the neual
netuok ttal
buld a
pnay wayto ditins for a
prorndes accuráte pre
Comnon hype
certan problem, noot
paramelrs incude the
toe actvatonfunthor
bhdden tayers,
-tmeolepchi)
and bo many
shauld Ge epcatd
tranng

Classic

You might also like