Scribd
Upload
19 views

Lecture 02 - Supervised Learning

1. The document describes a supervised learning task to predict whether a car is a family car or other type of car based on attributes like price, mileage, number of doors etc. 2. A model is trained on a dataset containing attributes of cars and their type (family/other). The model belongs to the family of rectangles and finds the best parameters to classify new data. 3. The trained model is used to predict the type (family/other) of new cars based on their attributes. The model parameters are updated iteratively to minimize errors in predictions.

Uploaded by

Ata
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
19 views

Lecture 02 - Supervised Learning

1. The document describes a supervised learning task to predict whether a car is a family car or other type of car based on attributes like price, mileage, number of doors etc. 2. A model is trained on a dataset containing attributes of cars and their type (family/other). The model belongs to the family of rectangles and finds the best parameters to classify new data. 3. The trained model is used to predict the type (family/other) of new cars based on their attributes. The model parameters are updated iteratively to minimize errors in predictions.

Uploaded by

Ata
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Supervised Learning

set
x
=

5(xi , yi) 3 training

whether is family cor

e
car

predicting
a

Task :
2

not xi tIR
~ in
=Faeoso
or
.

-
-

nee
2x1
if is a family cor
↳
!I
xi
: =
other wise
O

an
1x2 1st cer

↳Jean
X12
data
matrit
3
X = y
y
=

with car yN

- Nx2
NX1
family

asiy
car
⑧
⑧

type of
⑧

1
-

particl other
⑧
⑧ ④
car
⑧ &
·

A
.... *

22 ⑧ ⑧
O

(i)
⑧

0
31
=

x
⑧ X =
...
X 22
-

· 2

1
21
. . .& ⑧
y i
in
⑧
X12
-

↑2
,
price x =

( Y] 3 = 1

Model family
FAMILY OF RECTANGLES R = Ep1 ,
p2,
er
,
e23
* nu
ifp xm+ 1 <p e : Xr+1 , zen model
perometers
, -

YN+1
1 =

the best &


LEARNING =>
finding P 4 , 3
else JN+1 = 0 &
5
=2 p , ,

f(xn +1)p4 4 24 e2) , , ,


=
Jn+
Res
sor
resorti /]
-
warit(ii)
*# >X
X
-
vector

W
price X *
I

I
X
_
x1 =

cy X
. ·
....
......
-
31

i lita
32 X
-
-
-
x

#offeatures
32
,o
*
mileage(x> #of
detents
MODEL FAMILY
=
YMICLI TUES
,
LEARNING↓=> Ending the best

observed dicte M
*
Ew wo 3
estimated
=

get
,
We .
X +

W M

A
y
y , 31

:
M A er = -

WiXis+We
y
=

,
A 4
1
XN+ 1
22 =
↑
we X21+ wo y2 -32
Wo
=

42
.

! *
1
i Xia ,
+

I W

36-56
A
=
*
We .* 61+ We 26
=
YN+1 -

46
-
X minimize E(yi -i) -

T 1
ei lei

*
=

X minimize , ly- g)= i = I


leil
e:

w
ei
minimize Gisei Xe
:

minimize
"E(yi -

(1 .
Xi +

IX)
to We and we
with respect
.
:
Error(we ,
wel 4) = (y : -

(1xin +

wo))2
i = 1
-

2 Errou
-

2 wo
=
E (Etxin+wosl]
. 2 wo

I
0
=E 2 .
(yi (wixin
-
+

wo))( -
1) =

(wixil +web) (-xin)


* 0
2 Error
(y :
=

2 - .

-i
.

Owe 1

ii)
=

A
Solve for we

Exercise
N
N
-
N
/x1 yih
.

E -

As
.
.
-
-
T I
-
=

[E /N]
W1 - N

Ex
x:
-

we =

(-, 3: /N) -
w .

( xi/N) ,
e

5 -

You might also like