Improving Biometric System Privacy, Security and Accuracy
Terrance E. Boult
El Pomar Prof. of Communication and Computation, U Colorado at Colorado Springs
Founder and CEO, Securics,Inc.
October 11, 2005
Sage 3510 - 4:00 - 5:00 p.m.
Refreshments at 3:30 p.m.
The talk will start with a discussion of Biotopes, which are secure revocable
tokens computed from biometric tokens. This section of the talk will briefly
review the privacy/security issues with biometric, explain why standard
encryption does not solve the problems and quickly review the state of the art
in privacy preserving biometrics. It will then present the key ideas behind the
patent-pending Biotopes, and present performance results where existing face and
fingerprint algorithms were extended with Biotopes to improve security/privacy.
We discuss why the Biotopes and associated transforms, actually improved the
accuracy of the underlying algorithms. We conclude the segment on
privacy/security with some examples highlighting the important of accuracy and
why it has such a strong impact on privacy/security concerns.
The second aspect of the presentation will introduce our unique work looking at
predicting the failure of a biometric recognition system. The approach, Feature
Analysis of Similarity Surface Theory (FASST), conjectures that the the
similarity scores used in recognition contain information which can, in general,
predict when the system is failing. AdaBoost is used to combine the features
computed from the similarity surface to produce a patent pending system that
predicts the failure of a biometric system. Face-system Failure prediction,
using a leading leading commercial face recognition system, is presented as an
example to show how to use the approach. On outdoor weathered face data, the
system demonstrated the ability to predict 90% of the underlying facial
recognition system failures with only a 15% false alarm rate.
The final component of the talk presents our Random-Eyes(TM) approach, a novel
technique for improving face recognition performance by predicting system
failure, and, if necessary, perturbing eye coordinate inputs and re-predicting
failure as a means of selecting the a perturbation that provides correct
classification. This relies on a method that can accurately identify patterns
that can lead to more accurate classification, without modifying the
classification algorithm itself. Showing the generality of FASST, this time we
use a neural network trained on wavelet transforms of similarity score
distributions from an analysis of the gallery. Face images with a high
likelihood of having been incorrectly matched are reprocessed using perturbed
eye coordinate inputs, and the best results used to "correct" the initial
results. Results for both commercial and research face-based biometrics are
presented using both simulated and real data. The statistically significant
results show the strong potential for this to improve system performance,
especially with uncooperative subjects.
Dr. Boult the El Pomar Endowed Chair of Communications and Computations and Networking and Professor of Computer Science at the University of Colorado at Colorado Spring. As the El Pomar chair he works with the Colorado Institute for Technology Transfer and Implementation working with faculty and local companies to develop and transfer technology in the Springs area. In the 2003-2004 academic year Dr. Boult worked with local companies on over a dozen SBIR/STTR proposals.
Until July 2003 he was the Weiseman Chair Professor and the New Century Fund Professor of Computer Science in the Computer Science and Engineering Department at Lehigh University, where he was the founding chairman of the CSE department. From Aug 2002 through Jan 2004 he was the Chief Technology officer for GuardianSolutions Inc, helping to develop the next generation of sensor-based security systems.
Dr. Boult's ongoing research projects include advanced biometrics , advanced visual security systems, design and evaluation of imaging sensors for facial recognition, evaluation of weather effects on facial recognition, algorithms for efficient use of wireless networks, with funding by DARPA, ONR, the Army Night Vision Lab, and Siemens, the AT&T Foundation, the Lucent Foundation, RemoteReality Inc, McQ Associates, PadCom Inc, and Pennsylvania Infrastructure Technology Alliance and the Ben Franklin Foundation of PA.
Dr. Boult has published over 120 papers holds 6 patents. He has been the co-organizer of multiple workshops of Visual Surveillance, and computer vision related topics, and has been on program committees for more than a two dozen IEEE and SPIE conferences and workshops on computer vision and sensor fusion and as well on the ACM program committee in software engineering. He has served as an associate editor for IEEE Trans. on Pattern Analysis and Machine Perception (IEEE PAMI), Dr. Boult studied at Columbia University, earning his B.S. in Applied Mathematics in 1983, and his M.S. and Ph.D degrees in computer science in 1984 and 1986 respectively. He was on the faculty of Columbia University from 1986 until joining Lehigh in 1994. He received an NSF Presidential Young Investigator award, and has won teaching awards at both Columbia University and Lehigh University.
Hosted by: Bulent Yener (x6907)
For more information:
Terrance E. Boult
Last updated: October 7, 2005