Summi S: Dhara

Camlin Page
SYI T Date

Classiication Algontha in Mackin Learniag

Assign ment
Q.1) Rxplaih Superised learnihg algoitom
Tae sperised apprach d sìmi lan to huuman Leanaina
inder tne iuperisio a teacher The teacher amide
q0od exanpley for the atdent to memorizy tae itudn
then dei veA qeneral nles fram these spi7ie exanplss Hi
w se past data to ain ow machine < bayod n thot datc
lto traunomachine baye d on tnat data acwracuy or we can
Can be heck.

We teach machine,beueo on that e test the acuracy e

Machine eannina algantam an testiaq cdata for exane, to
ìnoplament aprce predi Cticn alaoitana based n machine
15
Leamingwe He have 2 year past data thi sto Cky. w wse
4 years data be uyed to tet 4 Qnalyie tae acacy Can
St come aecane we hane input dota 4 output data ca weu
as well tthati why its called superised leaznig
20

Q2 Lxplain unperdsed teanning algaithna

tai Leanniywwe do not tU et any particuda outcome
ke supen sed teasning Hene machine biu to gup dheent
obe cty based an the simila att ibwt the have, ce d on t
teanning input dota he outcom or the es t ttall
Here e have anl
dope nd an the intelligenci a the machine.hese are Called un
Aperi sed ceaning be caye unlika supervised (earning thene
Q3) Exelaih the hiatey Neural networe
Atiial eural Netueki ( a N a sìmlabin abshrotk
nelaro
uitcnm, wich oontains a collecUh o
cOnncioycalle
whiCh coninnilate eaclh OheY through
The ANN model hay a
dendiles in a nerrous stCm
wanppose
The st nmodel e) nett neural net Lwrks
Mc Ceulloch & pits int m o a conut otim ed
(443 by
o del neurol actihy
jallonwed ay cother pYo posed loy John VonnNea
The odel wcs Rosenblatt & many others
man Main Minkyfrank
functins d neuron
Broin hewYon
a.4)plain the concept ehmucte called denlit
n e m possess t boYel calledl
Many cein Slqha bom other neunn ia junctoy
wich
bu mean fes synapied
neuno comminicate
Bone
Other pese posses thoUsan.
functiy a neuron
brouh contan Over OO bjoh
tha e human n the
E U estincte e orec l000 ynapses
thakanury can have

Conds, but conneetii hundred

Uhe o) a few nmilise
Swith ng
heuYDm through t
tim higher rmaticn to other
neuron rasmut in
A
axoy hh depehdy on
dawmit elechicot ipuhei
) The axan
potntiali exei tat uns oy inibi faryt
tansmitted dota can be ra°ouy natn,hi
ran'oy
)The slgnals
30|
ree Ceives
ceres input,,
A neuro
ghmed
 [Camlin Prge
Date

aca'rated generate intor mati onal mestager

Ca.s)How classitcotian w0 rk?

we odit heter athing can be refemed to a partilar
Clos e very day To gii an exampleclaii7ication help Us
hale descision whenpicking tomatoes ta asupeYrmat kat
Cqeenperteet", "o tten)Th machiu learniny terms, we
asign a label og one tue classes o ererytomato we hold
noc hand.
h atticiantyoyauy Tanmat pickiy contut depnd nth
tn you qo h u upermar
kek yolEte bettey yau woiU become at picking -ot out
15
Tomatoes that ane freh&

Q) wit step to bitd ala siicatiea, mmodel

Once yau lnos whak kind o classiHeaten task you cdoaing
with,it iu time to buitd to modal
DSelect the classiierYou need to choose ahe
mthm that you wil appy t your dotas
2)TrGn t. ou hae to prepare ahaining deta &et with
lateled resuts
3)Pre dil the otput. Use tu mode to get som neult
4) aluat m classiie modelTt e aqood idea to prepae
avaidotim et a data taet you hare not Ae d in training
 Camlin| Page
Date

pitheentale betwe @n the Linee

mode.
innyS Logistic areyioh

linear Regreasion Logutic Regreioa

|to predit the Contim to preict the Caut egoricol
dpendent vaiabu wain lependet vaniablu wing
Gies t o independant agien sek a indapntnt
ioaabe ani cable
n)Ttu wsed for soling gre
siOn problumi iþatism pyoblu
3)ntee predict the 3)Tnthyene preo r the
volue dk Contous problim.
4)Tatheseae jind the heat 4)Tnthese, we ind stu curre
ft Lin by whih
oaaiy prdit th Output
Laest squas estimetin method S) Maimnm i kelihaad eatinnatio
mithod u yed or ebnatio
acuracy
20

6) he nelattons hp acuacy
hetean 6)Tt deDe not rquie a inear
independant relatiohip bteen depsndent
aicbla met ae ihear.
)heueEi collinecuuiey colineaia, batse
batween the indapentat
25

Vaiaoles.