0% found this document useful (0 votes)
7 views

Adobe Scan Mar 04, 2025

The document discusses various machine learning models and their approaches to estimating probabilities and conditional densities using Gaussian distributions. It covers concepts such as decision boundaries, covariance matrices, and the estimation of parameters for classification tasks. The focus is on how these models can be applied to classify data and the implications of different covariance structures on decision boundaries.

Uploaded by

Rohit Kumar
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
0% found this document useful (0 votes)
7 views

Adobe Scan Mar 04, 2025

The document discusses various machine learning models and their approaches to estimating probabilities and conditional densities using Gaussian distributions. It covers concepts such as decision boundaries, covariance matrices, and the estimation of parameters for classification tasks. The focus is on how these models can be applied to classify data and the implications of different covariance structures on decision boundaries.

Uploaded by

Rohit Kumar
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/ 7

03.o3.

2025
Greme ove Models
ML Abbo oach
dataset coniSts
baining data The
input Yaicble in

(o os L).
binasy dagdo

Cas Dtibuh on: The olato boino (rdo


am Coith
classes

Generahie Mo dels Thee modals do estinata the


P(n, Ce) uicl be esbresed
PCCA)
dmaiy
þ(Xn| Ce).
Cars

Psoss amd Condihion al Dersihes:


The pno oobab ilites fos cases C and C 7

’ The clas- Tepachvely


condih'onal densihies pln lCe)
Gasian distnbkons
Jont Dernsitie
’ for
for

Tho opfroacl
data coithin Gaessam distbafos Jo
Cac
o trobabiites dhad clas allorwing lestae esth
Carn wed
fage

Generahue Modulo - ML
Eshma
&male dhe
þaxamolens
he li7alihoo d henciem.
l:ltahood
Likaleboo funchon : lhe litalitood junchon s gem ta

aseere tha
aget loebel,

Conweient to z2
mani snize the
kikalihoock log p1|e). he loms mNolvirg
Jog- litalihooc

Paramets Estisnathion Seting tha desivabve e)

This inpies and


Plo]

Soluion C, an . Ce ase Cabulale


arnd
7he meas 0n eeclu cas
aerage of dhe dota poi nto

=
Page

The coiana malxi ces Galoelatil

Clas - Gondihenae Desithe CccD):


’ The esnaho o class
data som the other dass

’ This apþroaclu án eocvalent 1o


CCDs

Model Chasgct ecisties


’ The model
Tepsenengt ahon
desceipnie
the
fsomdig datailad
Clas
litbhon aoHin esc

dis coximÝnahu ,meanu


be tha best a nguis h g

HU - Shars .
Covaian ce Matix

s6enph¡n
Both Sha Coraian
Page

Ghod cotaelarne matoo


ML eshmale o

fos class

fox cfavs Cz

(an - 42)7 (an- e)

The Shased Covaiane malix ceeigted combinahi


the class matsice
Specfic.

N
t
am tha clase

Tespechiuely amd
the
2fter Conpuh g
The

Oecision Borday CShosad Covasciama)

boumday clase
obabiihes of quel
po steiot
e
P[C1l J : PI]
Date
Page

fh oam bo, Oxbrosad.

e[c JN( 144a , )

Hee,rlur, ) sopresemto ho. Aavsam


distibutin
Ge and ahoxod covarom.
matix

Becase the Co Varian onatoix


classe, the qaclaahio dernm the
distbeh on
Gamssion
The gesclts in a line a
dacision Goundoy 4

P[CL]
P[CI

blas fen.

The posieuot
the odbabuiy glven given

Thin stamoi d
o a fobabiuy Combia
Valees
Quadahie Deciston
Crunday
ovaite shaso
hedhe decision
boumda becom
Comlap, mO9-linoa
gwolrabhe ,
to

bounlon
dhe illusbah on decisiom
decisimoundasies
Ohared Covaeiana Malis
ohem und aseon) chaa
Clases_(e:g: Red amd
Same Covasiamee natsix thes dacisi on boundasy
b/w tharn

’ Thi linsasih avises becauuse the quadahie


the hausiam distibcuh ons Camcal ot, oexulbirg
line
bovrlasy
Btterent Covaiame Matuces
clas (eg. Blue) has diffeemt covaNane
Comforod to dhe hars, dhe daciioon

Suodabhie bne dacies Cusues eg: þouablas , hyperbolas)


Combe tharm iner bound

Decision Bonurdasies.
boumdaies Cam be
bninen decbim
POF dhat irnclyde highen -

moants allono yor exible amel


tligle oder bonndacies
im dhe docision
8hafesGaphoe sallens in tea.
more ndoieale potleins tle. ada
to
clASSM
Date
Page

a9lom one GcoLevs m

You might also like