|
28-28 Jun 2022 Villetaneuse (France)
|
Abstracts
_____________________________________________________________________________________________________ Mihai Mitrea Title: Visual content watermarking: between information theory and deep learning Abstract: Image/video watermarking emerged some two decades ago as a cross-disciplinary research field, combining principles from information theory, image processing and human visual perception. This way, the mark to be inserted stands for a message to be transmitted through an abstract channel where the original content itself and the attacks represent the noise sources; the emitter power is drastically limited by the human visual system sensitivity. Hence, both theoretical and methodological attempts have been made to mathematically model the information sources and the encoding constraints set to watermarking applications. Recently, the applicative performances of deep learning-based solutions imposed themselves as an opportunity to reconsider this theoretical model. The talk presents a succinct state-of-the-art survey on achievements related to both theoretical models and deep-learning usage for video watermarking. Rather than a confrontation, it tries to identify the synergy and the complementary among these two approaches and to identify the trends, as steamed for the needs none of these two approaches is ready to solve today. Biography Mihai Mitrea is currently an Associate Professor at Telecom SudParis, a school of Institut Polytechnique de Paris. He contributed to more than 20 European and French R&D collaborative projects, related to 2D/3D content protection and collaborative multimedia based application virtualization in cloud. He is/has been supervising 13 PhD thesis at Université Pierre et Marie Curie, Université ParisDecartes, MINES Paris-Tech and IP Paris. Inside ISO/IEC JTC1/SC29 (a.k.a MPEG/JPEG), he is an active contributor to the emerging standards related to multimedia content tracking and authentication, and he serves as an advisor for the French National Body. He is vice-president of the Cap Digital's Technical Commission on Digital Content. ______________________________________________________________________________________________________ Richard Jiang Title : Deep Facial Diagnosis: Sense Deeper toward Genotypes and Phenotypes Abstract : In this talk, we will review the recent advances on the study of the nexus between face and genetic/medical causes, and summarize the cross-disciplinary challenges and opportunities that can be worth of further efforts from the biometric communities. Biography Dr. Richard Jiang is a Senior Lecturer (Associate Professor) in the School of Computing & Communications at Lancaster University, UK. He is a Fellow of HEA and a Member of EPSRC College. His research interest mainly resides in the fields of AI Ethics, Private Learning, Quantum AI, Neuronal Computation, AI-Automated Healthcare and Satellite/Aerial Image Analysis. Dr Jiang's recent research has been supported by grants from EPSRC (EP/P009727/1), Leverhulme Trust (RF-2019-492), Qatar Science Foundation (NPRP No.8–140-2–065) and other industry/international funders. He has supervised and co-supervised 18 PhD students including 6 defended PhDs. He authored over 100 refereed publications and was the editor of a number of books and special issues. He has served as a PC/Editorial member and a reviewer for various international conferences and research journals. _______________________________________________________________________________________________________ Wassim Hamidouche Title: Federated Adversarial Training with Transformers
Abstract: Federated learning ( FL) has emerged to enable global model training over distributed clients’ data while preserving its privacy. However, the global trained model is vulnerable to the evasion attacks especially, the adversarial examples (AEs), carefully crafted samples to yield false classification. Adversarial training ( AT) is found to be the most promising approach against evasion attacks and it is widely studied for convolutional neural network (CNN). Recently, vision transformers have been found to be effective in many computer vision tasks. To the best of the authors’ knowledge, there is no work that studied the feasibility of AT in a FL process for vision transformers.
In this talk such feasibility is explored with different federated model aggregation methods and different vision transformer models with different tokenization and classification head techniques.
Biography Wassim Hamidouche received the master’s and Ph.D. degrees in image processing from the University of Poitiers, France, in 2007 and 2010, respectively. From 2011 to 2013, he was the Junior Scientist with the Video Coding Team, Canon Research Center, Rennes, France. He was a Postdoctoral Researcher with the VAADER Team, IETR, from April 2013 to August 2015, where he worked under collaborative project on HEVC video standardization. Since September 2015, he has been an Associate Professor with INSA Rennes and a Member of the VAADER Team, IETR Lab. He has joined the Advanced Media Content Lab of b<>com IRT Research Institute as an Academic Member in September 2017. He is the author/coauthor of more than 140 papers at journals and conferences in image processing, two MPEG standards, three patents, several MPEG contributions, public datasets, and open source software projects. His research interests include video coding and multimedia security. _______________________________________________________________________________________________________ Raja Kiran Title:
Abstract: The need to protect biometric data has been well advised according to various regulations and standards. The most popular Bloom Filter-based template protection schemes for iris recognition directly depend on the keys to avoid linkability challenges. This talk discusses existing approaches and a new approach for generating the keys directly from the iris biometric data using chaotic maps and elliptic curves over finite fields. The application of it will be discussed for template protection scheme that can directly exploit the generated keys to provide better security using a Quarter-Rounded template encoding which employs the inter-relation of bits in the neighborhood of the iriscode. Biography Raja Kiran received the Ph.D. degree in computer Science from the Norwegian University of Science and Technology, Norway, in 2016. He is Faculty Member with the Department of Computer Science at NTNU, Norway. He was/is participating in EU projects SOTAMD, iMARS, and other national projects. He is a member of European Association of Biometrics (EAB), chairs Academic Special Interest Group at EAB and is a senior member of IEEE. He serves as a program co-chair of BIOSIG. He is also a member of the editorial board for various journals. _______________________________________________________________________________________________________ Ahmed Bouridane Title : Imaging for Forensics and Security: From Theory to Practice Abstract : This talk provides a detailed analysis of new imaging and pattern recognition techniques for the understanding and deployment of biometrics and forensic techniques as practical solutions to increase security. It contains a collection of the recent advances in the technology ranging from theory, design, and implementation to performance evaluation of biometric and forensic systems. This book also contains new methods such as the multiscale approach, directional filter bank, and wavelet maxima for the development of practical solutions to biometric problems. Biography Ahmed Bouridane (Senior Member, IEEE) received an “Ingenieur d’Etat” degree in electronics from “Ecole Nationale Polytechnique” of Algiers (ENPA), Algeria, in 1982, an M.Phil. degree in electrical engineering (VLSI design for signal processing) from the University of Newcastle-Upon-Tyne, U.K., in 1988, and an Ph.D. degree in electrical engineering (computer vision) from the University of Nottingham, U.K., in 1992. From 1992 to 1994, he worked as a Research Developer in telesurveillance and access control applications. In 1994, he joined Queen’s University Belfast, Belfast, U.K., initially as Lecturer in computer architecture and image processing and later on he was promoted to Reader in Computer Science. In 2009, he joined Northumbria University at Newcastle leading the Computational Intelligence and Visual Computing Lab. His is now Professor of Machine Intelligence and Director of Cybersecurity and Data Analytics Research Center at the University of Sharjah, UAE. His research interests are in machine learning with applications to imaging for forensics and security, quantitative pathology and bimedical engineering, homeland security and video analytics. He has authored and co-authored more than 350 publications and two research books on imaging for forensics and security; and Biometric Security and privacy. |
| Online user: 2 | Privacy |
|
