You are on page 1of 28

Pearson VUE and IBG

Palm Vein Recognition for Candidate Authentication

Rubén Arturo Garcia Michael Thieme


Product Manager Director of Special Projects
Pearson VUE International Biometric Group
www.vue.com www.biometricgroup.com

Biometric Consortium 2008


Tampa, Florida
25 September 2008
Agenda
 Business case / rationale for biometric candidate
authentication
 Technology selection criteria
 Implementation status and future plans

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 2
www.biometricgroup.com Biometric Consortium
2 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 2
About Pearson VUE
 Part of Pearson plc, an $8 billion corporation that is the
largest commercial testing company and education
publisher in the world
 Delivers over 4 million high-stakes tests annually
 Licensure, certification, academic admissions, regulatory,
and government testing service markets
 5,200 test centers in 162 countries
– 230 fully-owned and -operated Pearson Professional Centers
– Pearson Professional Centers utilize a patent-winning design
created specifically for high-stakes testing

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 3
www.biometricgroup.com Biometric Consortium
3 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 3
About International Biometric Group
 45 employees
 Offices in New York, Washington DC, San Francisco
 Provide independent biometric services to government
and industry
 Business lines: integration, research, and consulting

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 4
www.biometricgroup.com Biometric Consortium
4 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 4
Pearson VUE Experience with Biometrics
 Pearson VUE an early adopter of commercial biometrics
 Deployed hundreds of fingerprint devices in 2000-2001
for candidate authentication
 Deployment was successful, but limitations emerged
– Difficult to upgrade / customize software
– Quality of enrollments became increasingly problematic
– Increased error rates
– No longer a competitive differentiator

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 5
www.biometricgroup.com Biometric Consortium
5 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 5
Pearson VUE and IBG
 Pearson VUE hired IBG in early 2007 to develop next-
generation biometric system
 IBG tasked with cost/benefit analysis, vendor /
technology selection, biometric system design /
integration, testing

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 6
www.biometricgroup.com Biometric Consortium
6 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 6
Biometric Applications for
Candidate Authentication

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 7
www.biometricgroup.com Biometric Consortium
7 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 7
One Workflow, Mutliple Applications
 Biometrics may perform multiple functions in candidate
authentication
Enrollment

Central Same-Day
Duplicate 1:1
Detection Matching

Central Return-
Watchlist Visit 1:1
Searches Matching

Local
Watchlist
Searches

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 8
www.biometricgroup.com Biometric Consortium
8 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 8
Candidate Enrollment
 Enrollment on first encounter with new system
 Enrollment is attended
– Operators trained in usage, presentation methods, feedback
 Throughput speed is important but negotiable
– Assuming that three presentations are necessary, enrollment
should be achievable within less than 20 seconds
 Enrollment forms the basis of all subsequent verification
and identification events

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 9
www.biometricgroup.com Biometric Consortium
9 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 9
Same--Day 1:1 Matching
Same
 After enrollment, candidates are authenticated on entry
into test labs
– Objective is to prevent “candidate-swapping”, increase test
integrity, allow monitor to focus on in-lab candidates
– Candidates may be re-authenticated several times per day
– Enrollment template stored to provide 1:1 matching; uploaded to
central database at end of test day (or periodically)
 Very few false non-match errors should be encountered
– Number of attempts, fallback approach to be determined
– Transaction time should be less than 3 seconds
– Security level should be “low”: don’t inconvenience candidates

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 10
www.biometricgroup.com Biometric Consortium
10 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 10
Return--Visit 1:1 Matching
Return
 Candidates may return days, weeks, or months after
enrollment for additional testing
– Objectives: provide candidates with a streamlined test
experience, ensure end-to-end integrity of candidate records
– Enrollment template retrieved from central database at beginning
of the test day
 More false non-matches are likely due to time lapse since
enrollment
– Number of attempts, fallback approach to be determined
– Transaction time should be less than 3 seconds
– Security level should be “low”: don’t inconvenience candidates
– Potential for re-enrollment depending on scenario (e.g. age of
enrollment)

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 11
www.biometricgroup.com Biometric Consortium
11 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 11
Local Watchlist Searches
 Pearson VUE encounters candidates who repeatedly
attempt to steal test items
– These candidates should be detected before testing to disrupt
item theft
– Searches should return results within roughly 30 seconds – exact
workflow is to be determined
 Local watchlist comprised of known or suspected
“thieves” may be searched on enrollment
– Watchlist distributed on a regional and/or test-specific basis
– In case of match, operator may issue test with items of little value
– Security level should be “high”: not necessary to find all fraud,
just a large majority of fraud

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 12
www.biometricgroup.com Biometric Consortium
12 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 12
Central Watchlist Searches
 Pearson VUE identifies candidates who attempt to proxy
test or establish multiple candidate identities
– Some unauthorized activities are detected through non-biometric
methods
– Searches can occur after testing is complete – days or weeks are
available to investigate suspected fraud
 Central watchlist comprised of tens or hundreds of
suspected “proxy” or “re-testers” will searched using
enrollment data
– Watchlist populated and managed on a regional and/or test-
specific basis
– Security level should be “high”: not necessary to find all fraud,
just a large majority of fraud

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 13
www.biometricgroup.com Biometric Consortium
13 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 13
Duplicate Detection
 Pearson VUE encounters fraudulent activities that can
only be detected through open database searches
 After filtering, databases on order of 150k-200k may
need to be searched
– Based on gender, geography, other demographic factors
– Searches can be conducted over hours
– Primary challenge to split and distribute database to facilitate
throughput
– Response time can be on the order of 16-20 hours
– Manage impact of false matches through fusion with face
recognition – exact implementation to be determined
– Security level should be “high” to reduce number of false
matches

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 14
www.biometricgroup.com Biometric Consortium
14 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 14
Technology Selection

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 15
www.biometricgroup.com Biometric Consortium
15 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 15
Factors in Technology Selection
 Form factor
– Deployed in constrained, desktop environment
 Privacy
– Deployed in countries in which there is considerable resistance to
fingerprint technology
 Usability
– Must accommodate reasonable variations in presentation
– Must be usable with little-to-no training or expertise

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 16
www.biometricgroup.com Biometric Consortium
16 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 16
Factors in Technology Selection
 Accuracy
– Emphasis on low false reject rate
 Cost
– Per-unit costs a major consideration with deployment of thousands of
devices
 Global support
– Deployment in Asia, Africa, Europe
 Differentiation
– Pearson VUE wanted to deploy a leading-edge technology for
competitive differentiation

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 17
www.biometricgroup.com Biometric Consortium
17 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 17
Technology Selection
 Vein recognition
– Fujitsu PalmSecure palm vein recognition offered the best
balance of functionality, performance, cost, and differentiation

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 18
www.biometricgroup.com Biometric Consortium
18 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 18
Technology Alternatives
 What technologies were considered?
 Fingerprint
– A logical alternative, but the privacy perception as well as the
lack of competitive differentiation were impediments
 Iris recognition
– Addresses accuracy, scalability, but form factor and costs made it
a non-starter at the time
 Face recognition
– A strong contender due to parallel webcam upgrade; many
models had face as a fallback or fusion technique; however,
unconstrained environment limits 1:1 utility

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 19
www.biometricgroup.com Biometric Consortium
19 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 19
PalmSecure Enrollment Rates (CBT6)

Fujitsu Sum: Transactions 1&2


Zero Palms Enrolled 1 0.08%
One Palm Enrolled 20 1.55%
Two Palms Enrolled 1269 98.37%
Total 1290 100.00%

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 20
www.biometricgroup.com Biometric Consortium
20 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 20
PalmSecure Accuracy (CBT6)

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 21
www.biometricgroup.com Biometric Consortium
21 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 21
Some Challenges
 Delivering 1:N functionality with 1:1 algorithm
– What is the most efficient way to spread matching across multiple
servers and support daily throughput?
 Housing
– Prototype housing presented challenges for larger- and smaller-
than-normal hands, leading to false non-matches
 Performance over time
– Testing indicates that false non-match rates tend to increase as
time since enrollment elapses; tied to the form factor issue
 SDK limitations
– Some creative use of enrollment and recognition data may be
necessary to meet all requirements

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 22
www.biometricgroup.com Biometric Consortium
22 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 22
Current Deployments and
Future Plans

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 23
www.biometricgroup.com Biometric Consortium
23 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 23
Current Status
 Pearson VUE launched a pilot program on August 19th
covering India and South Korea
– The pilot has been initially launched in India covering 11
permanent testing centers and several event/mobile testing
centers
– The pilot will encompasses both privately owned Pearson VUE
Testing Centers and third-part franchises
– Since launch we have enrolled over 2500 testing candidates and
have conducted over 25,000 1:1 verifications
 South Korea to begin on September 22nd
 Pilot results have been positive and we plan to move
forward with global implementation

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 24
www.biometricgroup.com Biometric Consortium
24 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 24
Future Plans
 October 2008
– Pearson VUE plans to launch the technology globally to over 162
countries, over 450 locations to be completed by Q2 2009
 Q2 2009
– Migrate Pearson VUE clients from fingerprint to Fujitsu
PalmSecure palm vein biometric technology
 Post Global Implementation
– Develop watch-list and duplicate detection functionality
– Offer other verification services with palm vein biometric
technology

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 25
www.biometricgroup.com Biometric Consortium
25 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 25
Lessons Learned
 Pilot launch date was tightly dependent on software
release dates and availability of new device form factor
 Did not anticipate immediate acceptance of technology
by internal users, clients and centers

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 26
www.biometricgroup.com Biometric Consortium
26 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 26
Conclusions
 Fujitsu’s PalmSecure palm vein biometric technology is
extremely accurate and easy to operate
 Pearson VUE’s interactive on-line training documentation
was sufficient for operating devices
 Enrolling and verifying candidates is easy to accomplish
 Significant reduction in enrollment and verification error
rates
 Candidates provided positive feedback in survey
information

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 27
www.biometricgroup.com Biometric Consortium
27 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 27
Thank you

Ruben Garcia
ruben.garcia@pearson.com

Michael Thieme
mthieme@biometricgroup.com

Pearson VUE and IBG: Deploying Vein Recognition for Candidate Authentication – 28
www.biometricgroup.com Biometric Consortium
28 2008 Biometrics 2006
CONFIDENTIAL © 2007 International Biometric Group 28

You might also like