universal and composite hypothesis testing via mismatched divergence jayakrishnan unnikrishnan lcav,...

47
Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli, University of Illinois Amit Surana, UTRC IPG seminar 2 March 2011

Upload: charlene-roberts

Post on 26-Dec-2015

214 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Universal and composite hypothesis testing via Mismatched Divergence

Jayakrishnan Unnikrishnan

LCAV, EPFL

CollaboratorsDayu Huang, Sean Meyn, Venu Veeravalli, University of Illinois

Amit Surana, UTRC

IPG seminar2 March 2011

Page 2: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Outline

• Universal Hypothesis Testing– Hoeffding test

• Problems with large alphabets

– Mismatched test• Dimensionality reduction• Improved performance

• Extensions– Composite null hypotheses– Model-fitting with outliers– Rate-distortion test– Source coding with training

• Conclusions2

Page 3: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Universal Hypothesis Testing

• Given a sequence of i.i.d. observations test the hypothesis

– Focus on finite alphabets i.e. PMFs

• Applications: anomaly detection, spam filtering etc.

3

1 2, , , nXX X

0 0

1 0

Null

Alternate : ,

: ~

~ unknowni

i

H X p

p ppH X

Page 4: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Sufficient statistic

• Empirical distribution:

– where denotes the number of times letter appears in

– is a random vector

4

1 2: , , , N

T

aa an

nn n

n np

n

an a1 2, , , nXX X

np

Page 5: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Hoeffding’s Universal Test

• Hoeffding test [1965]:

– Uses KL divergence between and as test statistic

5

0ˆ { ( ) }nH D p p I ‖

2N n

0{ : ( ) }q D q p ‖

0p

np 0p

Page 6: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Hoeffding’s Universal Test

• Hoeffding test is optimal in error-exponent sense:

– Sanov’s Theorem in Large Deviations implies

6

0ˆ { ( ) }nH D p p I ‖

2N n

0FA

*MD

ˆ( 0) exp( )

ˆ( 1) exp( ( ))

p

p n

p

p

H n

H

P

P

Page 7: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Hoeffding’s Universal Test

• Hoeffding test is optimal in error-exponent sense:

– Sanov’s Theorem in Large Deviations implies

• Better approximation of false alarm probability via– Weak convergence under

7

0ˆ { ( ) }nH D p p I ‖

2N n

0FA

*MD

ˆ( 0) exp( )

ˆ( 1) exp( ( ))

p

p n

p

p

H n

H

P

P

20 1

1( )

2n AD p pn ‖

0p

Page 8: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Error exponents are inaccurate

8Alphabet size, A = 20

Page 9: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Large Alphabet Regime

• Hoeffding test performs poorly for large (alphabet size)– suffers from high bias and variance

9

0

0

0

0 2

1)]

21

[ ( )]2

[ (p n

p n

Ap

nA

D p pn

D p

E

Var

2N n

A

Page 10: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Large Alphabet Regime

• Hoeffding test performs poorly for large (alphabet size)– suffers from high bias and variance

• A popular fix: Merging low probability bins

10

0

0

0

0 2

1)]

21

[ ( )]2

[ (p n

p n

Ap

nA

D p pn

D p

E

Var

2N n

A

Page 11: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Binning

11

Page 12: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Quantization

12

Page 13: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

General principle

• Dimensionality reduction

• Essentially we compromise on universality but improve performance against typical alternatives

• Generalization: parametric family for typical alternatives

13

{ }p

Page 14: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Hoeffding test

14

0p

np0( )nD p p‖

Page 15: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Mismatched test

15

0p

np{ }p

Page 16: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Mismatched test

16

0p

np

n̂p

{ }p

Page 17: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Mismatched test

17

0p

np

n̂p ˆ 0( )

nD p p

Page 18: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Mismatched test

18

0p

np

n̂p

0( )nD p p‖

0( )MMnD p p‖

Page 19: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Mismatched test

• Use mismatched divergence instead of KL divergence

– interpretable as a lower bound to KL divergence

• Idea in short: replace with ML estimate from i.e., it is a GLRT

19

0ˆ { ( ) }MM

nH D p p ‖I

np{ }p

ˆ0 0( ) ( )n

MMnD p p D p p

‖ ‖ML

Page 20: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Exponential family example

• Mismatched divergence is solution to a convex problem

20

01

( ) ( ) exp ( ) ( ) ,d

di i

i

x x fp p x

0( ) sup , ( )MMi i

i

D p p f p

Page 21: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Exponential family example

• Mismatched divergence is solution to a convex problem

• Binning when

21

01

( ) ( ) exp ( ) ( ) ,d

di i

i

x x fp p x

0( ) sup , ( )MMi i

i

D p p f p

( ) ( )iBi x xf I

Page 22: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Mismatched Test properties+ Addresses high variance issues

- However not universally optimal in error-exponent sense

+ Optimal when alternate distribution lies in • achieves same error exponents as Hoeffding• implies optimality of GLRT for composite hypotheses

22

0

0

0

0 2

)]2

[ ( )]

(

2

[ MMp n

MMp n

dp

nd

D p p

D p

n

E

Var

{ }p

where d

Page 23: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Performance comparison

23A = 19, n = 40

Page 24: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Weak convergence

• When observations

– Approximate thresholds for target false alarm

24

20

1( )

2MM

n dD p pn‖

0~ p

Page 25: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Weak convergence

• When observations

– Approximate thresholds for target false alarm

• When observations

– Approximate power of test25

20

1( )

2MM

n dD p pn‖

0~ p

0~ p p

20 0

1( ) ( ) (0, )MM MM

n pD p p D p pn

‖ ‖ N

Page 26: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

EXTENSIONSAND

APPLICATIONS

26

Page 27: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Composite null hypotheses

• Composite null hypotheses / model fitting

27

0

1

: ~ for any

~ , for any

,

: i

i

H X p p

q qH X

P _

P

P

Page 28: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Composite null hypotheses

• Composite null hypotheses / model fitting

28

0

1

: ~ for any

~ , for any

,

: i

i

H X p p

q qH X

P _

P

P

{ : inf ( ) }q q pD ‖

Page 29: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Composite null hypotheses

• Composite null hypotheses / model fitting

29

0

1

: ~ for any

~ , for any

,

: i

i

H X p p

q qH X

P _

P

P

{ : ( ) }q qD ‖ P

Page 30: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Weak convergence

• When observations

30

21

1( )

2n A dpn

D ‖ P

~ p P

Page 31: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Weak convergence

• When observations

• When observations

31

21

1( )

2n A dpn

D ‖ P

~ p P

~ p P21

( ) ( ) (0, )n pp pn

D D ‖ ‖P P N

Page 32: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Weak convergence

• When observations

• When observations

– Approximate thresholds for target false alarm– Approximate power of test– Study outlier effects

32

21

1( )

2n A dpn

D ‖ P

~ p P

~ p P21

( ) ( ) (0, )n pp pn

D D ‖ ‖P P N

Page 33: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Outliers in model-fitting

• Data corrupted by outliers or model-mismatch– Contamination mixture model

33

(1 ) p qp ò ò

Page 34: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Outliers in model-fitting

• Data corrupted by outliers or model-mismatch– Contamination mixture model

34

(1 ) p qp ò ò

Page 35: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Outliers in model-fitting

• Data corrupted by outliers or model-mismatch– Contamination mixture model

• Goodness of fit metric– Limiting behavior used to quantify the goodness of fit

35

(1 ) p qp ò ò

)( nD p ‖ P

Page 36: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Outliers in model-fitting

• Data corrupted by outliers or model-mismatch– Contamination mixture model

• Limiting behavior of goodness of fit metric changes

36

(1 ) p qp ò ò

21

1( )

2n A dpn

D ‖ P

21( ) ( ) (0, )n pp p

nD D P P N‖ ‖

Page 37: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Outliers in model-fitting

• Data corrupted by outliers or model-mismatch– Contamination mixture model

• Sensitivity of goodness of fit metric to outliers

37

(1 ) p qp ò ò

21) ( ) ( )

2( Tq p GD p q p P ò‖

2 2 ( ) ( )Tp q p G q p ò

Page 38: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Rate-distortion test

• Different generalization of binning– Rate-distortion optimal compression

• Test based on optimally compressed observations [P. Harremoës 09]

– Results on limiting distribution of test statistic

38

0ˆ { ( )( )( ) }nH pD p I ‖

Page 39: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Source coding with training

• A wants to encode and transmit source to B– Unknown distribution on known alphabet – Given training samples

39

~X pp

1, , nX X

Page 40: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Source coding with training

• A wants to encode and transmit source to B– Unknown distribution on known alphabet– Given training samples

• Choose codelengths based on empirical frequencies

40

~X pp

1, , nX X

log( ( ))x np x

Page 41: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Source coding with training

• A wants to encode and transmit source to B– Unknown distribution on known alphabet– Given training samples

• Choose codelengths based on empirical frequencies

• Expected excess codelength is chi-squared

41

~X pp

1, , nX X

log( ( ))x np x

21 1

1[ | ] ( )

2n

AX H pn E

Page 42: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

CLT vs LDP

• Empirical distribution (type) of

42

1

1( ) { }

n

n ii

x X xn

p

I

1{ }niX

Page 43: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

CLT vs LDP

• Empirical distribution (type) of

• Obeys LDP (Sanov’s theorem):

• Obeys CLT:

43

1

1( ) { }

n

n ii

x X xn

p

I

1{ }niX

{ ( )} exp( ( , ))p np N p n p ò òP

( ) (0, )n pn pp N

Page 44: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

CLT vs LDP

LDP• Good for large

deviations

• Approximates asymptotic slope of log-probability – Pre-exponential factor

may be significant

CLT• Good for moderate

deviations

• Approximates probability

44

Page 45: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Conclusions

– Error exponents do not tell the whole story• Not a good indicator of exact probability• Tests with identical error exponents can differ drastically over finite

samples

– Weak convergence results give better approximations than error exponents (LDPs)

– Compromising universality for performance improvement against typical alternatives

– Threshold selection, Outlier sensitivity, Source coding with training

45

Page 46: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

References• J. Unnikrishnan, D. Huang, S. Meyn, A. Surana, and V. V. Veeravalli,

“Universal and Composite Hypothesis Testing via Mismatched Divergence” IEEE Trans. Inf. Theory, to appear.

• J. Unnikrishnan, S. Meyn, and V. Veeravalli, “On Thresholds for Robust Goodness-of-Fit Tests” presented at IEEE Information Theory Workshop, Dublin, Aug. 2010.

• J. Unnikrishnan, “Model-fitting in the presence of outliers” submitted to ISIT 2011.

– available at http://lcavwww.epfl.ch/~unnikris/

46

Page 47: Universal and composite hypothesis testing via Mismatched Divergence Jayakrishnan Unnikrishnan LCAV, EPFL Collaborators Dayu Huang, Sean Meyn, Venu Veeravalli,

Thank You!

47