NSF Workshop on Fundamental - PowerPoint Presentation

Slide1

NSF Workshop on Fundamental Research Challenges for Trustworthy Biometrics:Panel Summary Slides 

Co-Chairs: Stephanie Schuckers, Clarkson UniversityLenore Zuck, National Science Foundation,Christopher Miles, Department of Homeland Security

This workshop is supported by the National Science

Foundation under Grant No. 1048975 .

Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

Organizing Committee:Karl Levitt, UC-Davis, Lawrence Hornak, WVU/NSF, Fabian Monrose, University of North Carolina-Chapel Hill, Bojan Cukic, West Virginia University, Susanne Wetzel, Stevens Institute of Technology Advisory Committee:

Michael Yura, BIMA/DoD, Jeff Dunn, NSA, Kelly Faddis, Scott Swann, FBI, Patrick Grother, NIST, John Savage, Department of State, Carl Landwehr, NSF, Michael Benardo, FDICSpecial Thanks to:

Workshop ChargeTo establish the fundamental research challenges for trustworthy biometric systems in identity management

MotivationTransactions (security, health, financial, social, etc) rely on establishment of trust Need for linkage between biological identity and our digital persona in order to establish trust Biometrics (measured identity) is one means to establish that linkage

Need to establish levels of trust Privacy preserving componentsProvides for anonymityScales in the information content (uniqueness)Scientific basis (aging, medical, etc.)Usability, performance, reliability, etc.

Need for Fundamental ResearchGovernment funding focused on biometric capture (sensors and systems)application oriented (e.g. border security, wartime detainment)concentrate on advanced signal and image processing techniques to add robustnessRecent reports highlight need:“National Strategy for Trusted Identities in Cyberspace”

National Academies Report on “Biometric Recognition: Challenges and Opportunities”  

Research NeedsAlgorithms/approaches/models that address privacy, acceptability, usability, and security of stored and transmitted biometric information Need to establish scientific fundamentals of identity science, as well as research at the intersection of identity management and cyberspace

Keynote Lectures and DiscussionSummary of the NRC Report “Biometric Recognition: Challenges and Opportunities”Lynette I. Millett, Senior Program Officer and Study Director, Computer Science and Telecommunications Board, National Research Council of the National

AcademiesBiometric Recognition’s Role in Identity ManagementDr. Elisa Bertino, Professor, Purdue UniversityAdvances in Molecular Processing and Analysis—Future Directions in Identity Science

Dr. James Landers

, Professor

, University of

Virginia

Dr. Joan

Bienvenue

, Chief Scientist, Lockheed Martin Corporation

Panel Talks and DiscussionFoundations in Biometrics—Privacy, Security and IdentityPanel Leads: Susanne Wetzel, Fabian Monrose Panelists: Colin Soutar, Nalini

Ratha, Adam Smith, Elaine NewtonFundamentals of Identity SciencePanel Leads: Chris Miles, Stephanie Schuckers Panelists: Peter Vallone, Arun RossFuture Directions in Identity SciencePanel Leads: Jeff Salyards, Jeff Palmer Panelists:

Jeff Salyards, Jeff Palmer, Brigitte Rolfe

Foundations in Biometrics—Privacy, Security and IdentityPanel Summary Slides

Panel Leads:

Dr. Fabian

Monrose

, UNC

Dr. Susanne Wetzel, Stevens Institute of Technology

Panelists:

Dr. Colin

Soutar

, CSC

Dr.

Nalini

Ratha

, IBM TJ Watson Research Center

Dr. Adam

Smith,

Pennsylvania State University

Dr. Elaine Newton, NIST

Privacy, Security and IdentityPanel Summary Slides

Develop and assess systems which preserve secrecy/privacy of biometricse.g., biometric cryptosystems,

biometrics as keys, template protection, revokable biometrics, etc.

Taxonomy

of

systems

Evaluation of

strengths/vulnerabilities

of stored biometric

templates

Modeling data and

attacks

Metrics

to evaluate

systems

Development of advanced methods

Theoretical models of classes of similarity/distance measures beyond vector norms (e.g., for minutiae, face images, ...)

Error-correction techniques for complex similarity/distance measures

Apply

homomorphic

encryption to

biometrics

Anonymous

search

capabilities

“No

effort”

re-enrollment

Develop and assess systems which preserve secrecy/privacy of biometrics (continued)

Hardware: extending the biometric matching capability in trusted environment such as secure hardware co-processors, smart cardsQuantify the use of biometrics on diverse platforms, such as cell phones and

laptops

Quantify use of

biometrics within

multi-factor assurance

context

Q

uantifiable

and commensurable biometric strength of

authentication when combined

with other factors

Develop methodologies for

end-to-end security

evaluations

Define

evaluation methods for

anti-spoofing/

liveness

testing for each major modalityPrivacy, Security and IdentityPanel Summary Slides

Strengthen collaboration between different disciplines (cryptography, signal processing, security)

Harmonize language and terminology

Need for common large data sets, well-defined challenge problems

Encourage cross-disciplinary events/funding opportunities

D

esign

systems subject to

privacy constraints, public perceptions, and legal regulations

– incorporate legitimate concerns

How to responsibly collect/store/share operational or research data

Design to limit

function

creep

, clarify

functional

requirements

Education

on biometrics

benefit

society, privacy laws, etc.Privacy, Security and IdentityPanel Summary Slides

Fundamentals in Identity SciencePanel Summary Slides

Panel Leads:

Chris Miles, Department of Homeland Security

Dr. Stephanie

Schuckers

, Clarkson University

Panelists:

Dr. Peter

Vallone

, National Institute of Standards and Technology

Dr.

Arun

Ross, West Virginia University

Develop models for individuality (uniqueness)Biological modelsTo understand the biological basis of a trait; level of detail observed (e.g., Level I, II or III for fingerprint)Feature and score modelsTo utilize feature

and score distributions to deduce capacity of a template, or information entropyError Rate modelsTo model error rates based on empirical evaluation of algorithms; understand the notion of uncertaintyEstablish basis to assess impact of age on biometricsMethods to estimate biological age of an individual from biometric

Models to understand how biometric traits age with time; impact of disease on biometric traits

Face, Iris, Fingerprint

Automatic

template aging

and

update

schemes

Fundamentals in Identity Science

Panel Summary Slides

Understand impact of fusion on scalability and securityMethods to increase the capacity of a templateEstimating scalability of individual biometric traitsMethods to incorporate adversary attacks

in a biometric fusion frameworkFundamentals in Identity SciencePanel Summary Slides

Future Directions in Identity SciencePanel Summary Slides

Panel Leads:

Dr. Jeff

Salyards

, US Army Criminal Investigation Laboratory

Dr. Jeff Palmer, MIT-Lincoln Laboratory

Panelists:

Dr. Jeff

Salyards

, US Army Criminal Investigation Laboratory

Dr. Jeff Palmer, MIT-Lincoln Laboratory

Brigitte Rolfe,

MITRE

Extend methods for validation of forensic scienceMeasures of uncertainty, accuracy, reliability, treat as “pattern recognition”, relationship with biometric performance metricsPeer reviewed research on human observer biasExpand definition usage of biometrics beyond “recognition”

Intent? Medical? Group? Levels of identity?Establish science for assessment of new/novel biometricse.g. odor, cardiovascular, vein, etc., Extend application spacesStand-off, unconstrained, non-cooperative, covert, etc.

Future Directions in Identity Science

Panel Summary Slides

Advance rapid DNA as a biometrice.g. automated rapid DNA systems based on digital microfluidics, bio-material scienceSystem improvementsReduce cost and

time, automate, increase portabilityImprove sample methods of DNA, best recovery, non-invasiveNew markers Automated analysis and searching algorithms, secure and private communications, database managementScientific understanding

Effect of agingCountermeasures

Broader considerations

Identify applications (kinship, identity, other)

Keep

DNA

‘

in the loop’ with established biometrics, lessons can be learned from both

areas

DNA

training, tactics,

procedures

Fundamentals and Future Directions

in Identity Science

Panel

Summary

Slides

From Both Panels