Dr. Sejun Song

Facial Recognition Systems (FRS) have witnessed remarkable accuracy, robustness, and reliability improvements owing to Machine Learning (ML) advancements, rendering them highly viable biometric identity verification solutions across diverse authentication applications. However, concerns surrounding privacy violations, fake presentations, potential biases, and security vulnerabilities have arisen with the widespread adoption of FRS. In particular, the emergence of sophisticated AI-driven manipulation techniques, such as DeepFake, poses critical threats to security and privacy, enabling the propagation of misinformation and deception for identity theft.

This presentation delves into cutting-edge technologies that address challenges faced by FRS, particularly in the context of DeepFakes and identity theft. To combat facial biometric Presentation Attacks (PAs), liveness verification, and Master Face Dictionary Attacks (MFDA), we propose intelligent ML-based specular highlights detection methods. We postulate that existing facial spoofing techniques fail to accurately replicate the reflective components in counterfeit presentations, thus offering avenues for detection. Firstly, we introduce "Apple in My Eyes" (AIME), a software-based facial liveness detection method that efficiently detects PAs with high accuracy (~200 ms) without the need for additional costly sensors or user involvement in active responses. Secondly, we present novel ML-based DeepFake detection technologies, including "CHIEFS" (Corneal-Specular Highlights Imaging for Enhancing Fake-Face Spotter), "MobiDeep" (Mobile DeepFake Detection through ML-based Corneal-Specular Backscattering), and "READFake" (Reflection and Environment-Aware DeepFake). CHIEFS leverages corneal-specular and facial highlights features, combined with environmental factors, achieving improved detection accuracy of up to 99.00% using the ResNet-50-V2 architecture. MobiDeep, a real-time, cloudless, and lightweight mobile application, detects sophisticated DeepFake images with 98.7% accuracy within 200 ms. Meanwhile, READFake utilizes specular highlights on various facial and body parts and environmental factors, surpassing state-of-the-art methods with a 99.0% accuracy rate. Finally, we present a novel countermeasure against Dictionary Attacks using Reflection-based Identification (DARI) system. Leveraging a lightweight and low-latency vision transformer, DARI identifies inconsistencies between facial image specular highlights and physiological characteristics, achieving high detection accuracy (97.83% to 99.56%) against public GAN-face detection datasets, with instantaneous detection speed (less than 11 ms) against SOTA master face dictionary attacks. These proposed ML-based techniques offer robust safeguards against DeepFakes and identity theft, advancing the security and trustworthiness of facial recognition systems in contemporary authentication scenarios.

Biography:

Dr. Sejun Song is an Associate Professor in Computer Science and Electrical Engineering at the University of Missouri – Kansas City (UMKC). Song conducts Cybersecurity, Machine Learning, Internet-of-Things, and Mobile and Wireless Communications research projects for transportation, smart agriculture, smart cities, and eHealth applications. Song received his Ph.D. in Computer Science and Engineering from the University of Minnesota, Twin Cities, in 2001. He worked for Texas A&M University, College Station (TAMU), as an assistant professor and a founding director of the Cisco Test Engineering Center (Cisco-TEC). Before joining academia, Song worked in industries including Cisco Systems and Honeywell Research Lab. He is a recipient of the Center for Teaching Excellence Scholar for excellence in undergraduate teaching, a Faculty Teaching Excellence Award, a NASA Summer Fellowship Award, Air Force Research Lab's Visiting Faculty Research Fellowship Awards, a Cisco Summer Fellowship Award, and several best research video/paper awards including CCNC 2019, ISC2 2018, Mobisys 2014, ICCCN 2014, and CIEC 2013. Several agencies have funded his research, including NSF, AFOSR, Cisco Systems, ETRI, NIST, AFRL, NASA, NIH, CDC, TAMU, KT Research, and MoDoT.

Address:United States

Mr. Joseph Verardo

The Computer Science Department and the Electrical and Computer Engineering Department at Binghamton University have received a five-year, $3.5 million grant from the National Science Foundation's CyberCorps: Scholarship for Service (SFS) program to recruit and train talented students for entry into the government workforce and long-term success in cybersecurity careers. 

Each SFS scholar will be funded for two to three years. Upon graduation, the scholar will work in a cybersecurity-related position in a government organization, a national lab or an SFS institute for a period equal to the length of the scholarship. 

Each scholar will have a faculty advisor who will mentor the scholar and involve the scholar in a cybersecurity-related project through independent studies, MS project/thesis or PhD dissertation research. Each scholar will undergo a rigorous curriculum, research, and practical training through cybersecurity courses, personalized academic and research advising, cybersecurity conferences/seminars, Capture The Flag Club and cybersecurity competitions, and outreach activities. In addition, scholars will be trained in critical skills needed to succeed in their careers, including ethics, communication, writing, presentation, critical thinking, and teamwork. 

In this talk, Mr. Verardo will cover multiple questions that are often asked by applicants: What is covered? What is the service commitment? How to apply?

Biography:

Joseph Verardo's work includes promoting educational opportunities and advocating for the betterment of education across the United States. He is the inaugural coordinator of the National Science Foundation's CyberCorps Scholarship for Service program at SUNY RF at Binghamton University. Joseph served as an appointee on a committee at the United States Department of Education. While on the committee, Joe successfully proposed policies advancing the interests of students and educators. Joseph previously served as Vice President & COO and Director of Finance & CFO of the National Association of Graduate-Professional Students, as a trustee on the Stony Brook Council at Stony Brook University, Assistant Head of College Engagement at JerseySTEM, and as a Capitol Hill engagement consultant. While with the NAGPS Joseph advocated for higher education to Congress and the executive branch, and organized national conferences focused on leadership development and legislative affairs. At Stony Brook University, Joseph worked on initiatives aimed at increasing financial opportunities, enhancing student life and student representation in university affairs.