Scribd
Upload
3 views

I239-3-handout

The document appears to contain a series of fragmented data and calculations, possibly related to a machine learning model's training and evaluation metrics. It includes references to confusion matrices, training and test sets, and various statistical formulas. The overall structure and content are disorganized, making it difficult to extract coherent information.

Uploaded by

anhnguyenle0115
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
3 views

I239-3-handout

The document appears to contain a series of fragmented data and calculations, possibly related to a machine learning model's training and evaluation metrics. It includes references to confusion matrices, training and test sets, and various statistical formulas. The overall structure and content are disorganized, making it difficult to extract coherent information.

Uploaded by

anhnguyenle0115
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/ 4

12/10/18

92 ( 323 3 1 0
-2 2 3 C - 01/ 11
92 ( 323 C 0 A
C 1 11
C /1 / 2
C 1 2
) -123 ( 323 C ?
C A 8 1 0

/? 1: :0 8 820 8 1 2C 2 4 ? /A
( D4C E 2A 8 C2
H 8C 0 C ? 382 8 20 ?E
0 24C C? 2 8 A2 4? 4 D ?
G 0 : ? 382 0 8 0? E
0 CA2 8 2C2 C? D 2 /
H 0 ?E 0 ? 0 F
G 0 : C 8 0 ? 0 C / C? A 4C 42C 8?A ?A E C
H 0 ? , 8 2 0 3 3 20 8
H ,8 ,8 ,8 2 : E C

/A2 8 E C
4 ? /A ?
2 )2C 8?A

)2C 8?A

/0 28

/ H G E
E
I5 , A 5D8 4
8 5E A AA
/ / :C G:8 C 0C 8EE
1AAE E E , A
1CI 2 D8 G8 E: CC 8
AH
88 1CI E: CC 8
1CI D8 G8 E: CC 8
5 E 8 5E A 5 D8 G8 E: CC 8
AA C58 5E
1CI 1CI D8 G8 E: CC 8
G C E 2 AA
6 H DE J 8
5 E 5 1CI 2 6 H DE J 8

E A8 : 5E C C 0A G5 A: 2 6 H DE J 8
E : 5E C
5: A8 5 L 78 G E E CG E:D

05 38

1
12/10/18

8GH GNF I F FA G 8 0:-20 :8 A


Q ? LI M? I GF FA - 8 01 - 1 - 8
-F P ? R ? G F DD IGG 8? ?0 10- :0 0 -:- 0 0 0 - 8:/ 2
7HD GF IGG ? 8 0: - 1 - 8
?H? LF D 01 18: - 8 - 8
• DD FG ? I? D ? – " # = −&'()*(#)
• 3G EGI? ? LI? G ?
:8 A 0 0/ - 0 81 18: - 8 - 8
• L? GF : P ? R HD – - = − ∑ / # &'()/(#) H

/? A 0F GE? L GF 0:-20 :8 A
/A 1GN /A 1GN /A .I L ? – -0 = − ∑ *(1) ∑ / # &'()/(#)
GGD P(x)
GGD
2 D? F M?I
A? -20

: 4 29 83 /3756
- 7 - 3 6
/ 4 6
/756 0 :8
/3756 / 4 6 0
1:55 6
0 - - - - E :8 , , ,
– "# = − '⁄( )*+, '⁄( − ,⁄( )*+, ,⁄( = 0.92123 ' ' + +
– "# = − ⁄' ()*' ⁄' − ⁄' ()*' ⁄' = 0-./
/756- - - - - +- E 1:55 6, , , +,
– "4 = − (⁄5 )*+, (⁄5 − ,⁄5 )*+, ,⁄5 = 0.97bit – "0 = − 1⁄2 ()*' 1⁄2 − '⁄2 ()*' '⁄2 = 0.92bit
E0 , , , ,
• " ; = (⁄< ∗ 0.92 + 5⁄< ∗ 0.97 = 0.95bit – "9 = − +⁄2 ()*' +⁄2 − 2⁄2 ()*' 2⁄2 = 0-./

• " ; = '⁄< ∗ 0 + 2⁄< ∗ 0.92 + 2⁄< ∗ 0 = 0.35bit

1253 , + 2386 +

- E6 - 2 :5 -4 32 01 ,1 1/ 8
/ 78 24 2 5
38 :
G -4 32 38 :
1 5 E
• ,1 1/ 1 1 1 18
24 2 5 38 :, , , ,
– "′$%&'() = 0.95/01
– "# = − '⁄( )*+' '⁄( − '⁄( )*+' '⁄( = 1-./
– "′23456% = 0.35/01
1 5 E, , , +,
– "′89:4;)&53 = 0.85/01
– "0 = − 1⁄2 )*+' 1⁄2 − '⁄2 )*+' '⁄2 = 0.92bit
/ 78 38 :, , 80 A 1 1 2 1 / 8 01
00 1 2 8/ 1
– ": = − 1⁄1 )*+' 1⁄1 = 0-./
– "′$ %&' ( ) = 0.67/01
– "′8 9 : 4; )&5 3 = 0/01
• " < = (⁄= ∗ 1 + 2⁄= ∗ 0.92 + 1⁄= ∗ 0 = 0.85bit

0275 + 1

2
12/10/18

) 0 02A 8 A82 / 2 4
• (3C0 A0 • . 1 E 2 . 00 . 4
A8 8 8C 0 3 0 A A 2 A 2A
• C 2 2 . 2 . 2 2 / 1 22
A 3 38 A0 2 2A8 08. 0.
- 0 8 30A0
• 2 2 .8 2 . 2 0 20 /20. 2
A8 0 A 3 8C 0 A /
• 2 1. . .0 02
I 8 03C0 A0 • 2 2 02 4 . 2
A D8A 0 20 0 3 0 30A0
• 8. 0. C 4 /20. 2 22
A 8 2A0 0 0 8 0A8 A 0A 8 A A
1 3 30A0 A • 282 . . / 2 0. 2 8 2 4
A 0 3 0 C0 3 2 A8 0 0 A 2 . 4 1. .
0 2 A 0 1 8 A 8 81

0 .42

51 80 0880 / 6 4 206 6 22
88 0 8
8 5 80
• 6 4 6 1 2 / 2 8.06 4 . 82
/ 22 / . 82. 12
• 2 8.06 4 . 2 8.02 6 . 82 2 ./86 2
. 2 2 20 21 2 .26 2 / 22
0880 / 6 4 2. 2 . 6 2 6 482 82.
2 11 8 .6 6 4 2 6 4

264 264 . C
64 64
0. 0.
88 0 8 . C . 0. C
0. C
/ .42

1 0/ 0 ? / 2 / 1/ 8
• A part of training set can be used to • Confusion Matrix on Test Set
development (check overfitting) / 21 2
/8 8 /8 8
0 / /8 8 2 2 / 2 2 / 2
Training set 2 /8 8 / 2 2 2 2 / 2
Training set
)*+),
Development – "##$%&#' =
set )*+-*+),+-,
)*
– ./#&00 =
Test set Test set )*+-,

/ 2

3
12/10/18

80 ?0 20 9 CC 23 3D
, 3 03 D CD D D C3 3D3
• Confusion Matrix on Test Set - D 3 CD3 8 F 8 DD
80 2 CC 23 3D
09 9 09A 9 / C F 3 C3 3D3 C D 3C D D CD C D

1 0 09 9 / ? 0 0 ? 0 3 D C D 3 3 D 8D C D
09A 9 0 ? / 0 ? 1C D D CD C D D 3C C 8 3
• 8 CC F3 3D
+,
– "#$%&'&() = • /3 C DD 3D3C D D 8 C
+,-.,
• 3 8 3D3C D
4567899:,;67<=<>?
– / − 1$2'3#$ = 8 C 8D 3D3C D
@ABCDD - EFABGHGIJ • 0 D CD D D D 88 DF CC
• / 3D D C D 3 8D 8 C

0 .3

88 1 1 /- .L G D 8A :8H DC
18 1= 1 .2 . 1 1 0- 2 1- 0 H 1C D .:I 8H DC 2D C ,A8GG
3D 4 ::A CJ G HM ,8C
0 0 CJ G HM ,8C
0 3D 0 G DDA ,8C H
3D 0 / 8:I8H G DDA ,8C
0 4 ::A / 8:I8H G DDA ,8C H

48 DC IG DC 8H L
CG 8 I8 M 8AA E G DC 8GI

/- 58

- 4 G N B 8 6 D8 A 3
• 1 P E C B BG G .4 8 3 A - 4D E 4 : D8E
0 G L 28 86E AAE A 8 D : 4 8 4:8 8 E A I
-B HD 8 -B HD 8 2 E 4E4 A AAE E8DE
ID 5 H GDG L ID 5 H GDG L
E - E -
18 84E E
4HG 9 8 B 9 4HG 2 8 B 9
• A 8D 4 8 6 4DD8D
467 9 8 B 9 4 9 8 B 9
• /A A 8 84E 8D EA E8DE
4HG 2 / 2 4 2 / 9
467 9 / 2 467 9 / 2
18 68 8 A : I D : -
467 9 8 B 2 4HG 2 / 9 4D 8 8 A 8 E D8E
467 9 / 9 4HG 2 / 9 18 84E EA 8 468 D E 88 EA ADE 4 84 8
467 9 8 B 9 467 2 8 B 2 6 4DD 4 E 8 A 8D 4 6A 4 8 AE8 E 4
4 2 8 B 2 4 9 / 2

3 04:8

You might also like