large scale applications

15
Large Scale Applications Dr. Pushkin Kachroo

Upload: jack-sweeney

Post on 01-Jan-2016

39 views

Category:

Documents


6 download

DESCRIPTION

Large Scale Applications. Dr. Pushkin Kachroo. TexPoint fonts used in EMF. Read the TexPoint manual before you delete this box.: A A. Large Scale Applications. To reduce # of matches needed Filtering : using non-bio like sex, age etc. Binning : using additional biometric. Size. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Large Scale Applications

Large Scale Applications

Dr. Pushkin Kachroo

Page 2: Large Scale Applications

Large Scale Applications

• To reduce # of matches needed– Filtering: using non-bio like sex, age

etc.– Binning: using additional biometric

Page 3: Large Scale Applications

Size

• Small: 1000 (approx. stored samples)

• Medium: 10K• Large: 100K• Very Large: 1M• Extremely Large: 10M

Page 4: Large Scale Applications

Examples

Voter Registration

Enrollment Authentication

National ID

EnrollmentAuthentication

Page 5: Large Scale Applications

FingerprintFAR(1) FRR

1 Finger 10-5 10-3

2 Fingers 10-10 10-3

Page 6: Large Scale Applications

Enrollment False Positives

• Quadratic ~ [m x FAR(1)] x m• For one finger…10^8 gives 10^11• Two finger it gives 10^6

Page 7: Large Scale Applications

Enrollment False Negatives

• FNR ~ 10-3

Page 8: Large Scale Applications

Authentication Accuracy

• False Rejects: FRR x m = 105 for 108 size

• False Accepts: 10-5

Page 9: Large Scale Applications

Matcher Sizing

• m(t): Database size• J(t): Match throughput; #matches

needed/time• E(t): Enrollment Rate• A(t): Authentication Rate

m(t)

Page 10: Large Scale Applications

Matcher Throughput

tEdEtmt

avg 0

)()(

tEtmtEtJ avge2)()()(

avga AtJ )(

Page 11: Large Scale Applications

Matcher Throughput-2

m(t)

0

02

)()()(ttmEA

tttEAtJtJtJ

avgavg

avgavgea

)(tJ

t0

National ID

st

)(tJ

t0

Voter Registration/Election

Page 12: Large Scale Applications

Total # of Matches

mtJdtJdttJ aavg

t

t

avg

t

t

aa

00

)(

Verification on Election Day

Enrollment

)(22

)()( 2

22

0

2

0

mOmtE

tdtEdttJ eavgt

t

avg

t

t

ee

Page 13: Large Scale Applications

Cost

• Incremental Cost : Sampling, printing ID, etc.; Linear

• Exception Handling : Cost of Resolving False Positives; quadratic

m(t)

mihmh avg )(

Page 14: Large Scale Applications

Total Cost

m(t)

32

2

)1()1(

)()1(

)(])1([1)(

miFARmFARhm

mmihFARm

mhmFARmmmCost

avg

avg

Even for i = 0 we get:

12

1

)()1()1(

)()1()(

mhFARifmFARh

mhFARifmmCost

avgavg

avg

Page 15: Large Scale Applications

Cost-Plot

m(t)

)1(

FARhm

avg

Cost(m)

m