Biometrics in an Adversarial Environment: Security and Privacy
Abstract: Biometrics is the science of recognizing individuals based on their biological and behavioral attributes such as face, fingerprints, iris or gait. In its simplest form, a biometric system compares two biometric samples, such as a pair of face images, and determines if they originated from the same person or from two different individuals. Such systems have been incorporated into a number of access control, law enforcement, border security and consumer electronic applications for person recognition.
Notwithstanding the tremendous progress made in biometrics over the past decade, these systems are vulnerable to a number of attacks. For example, a face mask or a fake finger or even a printed iris image can be used to “spoof” a biometric trait. Similarly, an adversary can use facial makeup or cosmetic contact lenses to “hide” their real identity from a face or iris biometric system. Such type of “presentation attacks” can evolve over time and newer, previously unseen attacks can be launched against the system. With the incorporation of biometric sensors in personal smartphones as well as border security systems, it is critical to develop robust techniques that can detect and deflect such known and unknown attacks.
Even as biometric researchers grapple with vulnerabilities associated with the technology, the issue of personal privacy is being brought to the forefront due to the deployment of biometric systems in public spaces such as shopping malls where an individual’s identity can be surreptitiously unmasked. Further, there is concern that biometric data provided by an individual for a specific purpose (e.g., access control) will be used for other previously unexpressed purposes (e.g. demographic profiling). Thus, there is an urgent need to develop techniques that can impart privacy to biometric data as well as ensure that data collected for one purpose is not covertly used for other purposes.
In this tutorial, we will first introduce the fundamentals of biometrics. Next we will describe the vulnerabilities of a biometric system, including presentation attacks. Then we will discuss techniques that have been developed to counteract these vulnerabilities and attacks. Finally, we will describe some of the privacy concerns associated with biometrics, and present techniques that can be used to impart privacy to an individual’s biometric data.
Bio: Arun Ross is a Professor in the Department of Computer Science and Engineering at Michigan State University. Prior to joining the MSU Faculty in January 2013, he was with West Virginia University (WVU) as an Assistant Professor from 2003 to 2008 and as an Associate Professor from 2008 to 2012. He also served as the Assistant Site Director of the NSF Center for Identification Technology and Research between 2010 and 2012. His research interests include pattern recognition, classifier fusion, machine learning, computer vision, and biometrics. He is the coauthor of the textbook “Introduction to Biometrics” and the monograph “Handbook of Multibiometrics”, and the co-editor of “Handbook of Biometrics”. He is a recipient of the IAPR JK Aggarwal Prize, the IAPR Young Biometrics Investigator Award, the NSF CAREER Award, and was an invited speaker at the Frontiers of Science Symposium organized by the National Academy of Sciences in November 2006. He is also a recipient of the 2005 Biennial Pattern Recognition Journal Best Paper Award and the Five Year Highly Cited BTAS 2009 Paper Award.