Special Sessions

BTAS 2016 will feature the following Special Sessions. The feasibility of a Special Session will be based on the number of accepted papers related to the topic of the Special Session.

Important: Authors who would like their papers to be considered in one or more Special Sessions, must select the appropriate Primary and/or Secondary Subject Areas when submitting their paper.

  • Biometrics and Multimedia Forensics
  • Biometrics on Mobile and Wearable Devices
  • Face Spoofing Detection in Real-World Applications
  • Ocular Biometrics in the Visible Spectrum
  • Representation Learning and Biometrics

Biometrics and Multimedia Forensics

Organizers: Chang-Tsun Li, Warwick University, UK; Tieniu Tan, Institute of Automation, CASIA, China; Massimo Tistarelli, University of Sassari, Italy

Multimedia forensics is concerned with the development of scientific methods to extract, analyse and categorise digital evidence derived from multimedia sources, such as imaging devices. For example, developing technologies to identify, categorise and classify the source of images and video, as well as to authenticate and verify the integrity of their content. The enabling technologies in multimedia forensics are quite similar to those used for identification and verification purposes in biometrics.

In the past few years many research laboratories have successfully explored the application of biometric technologies to problems pertaining multimedia forensics. This line of research resulted in a number of outstanding scientific works which rapidly led to the development of practical tools as an aid for criminal investigations. Several research funding agencies across the globe have given a substantial support to research efforts in this field. Among them the European Commission recently funded the project IDENTITY, while the US National Science Foundation also launched a new program for funding research on the same areas.

Given the growing interest, but also the remarkable achievements in the field, we propose to organize a special session at BTAS 2016, specifically devoted to present high quality research papers combining biometric technologies and multimedia forensics. The contributing papers will cover different modalities, approaches and application scenarios. This special session will collate the efforts and major achievements of scientists working in the field, in an attempt to contribute to robust solutions to the many challenging problems related to this emerging research area.

Biometrics on Mobile and Wearable Devices

Organizers: Vishal M. Patel, Rutgers University, USA; Kiran Balagani, New York Institute of Technology, USA; Paolo Gasti, New York Institute of Technology, USA

Recent developments in sensing and communication technologies have led to an explosion in the use of mobile and wearable devices, such as smartphones, tablets, and smart watches. These devices routinely collect, store, and process personal information such as health data, location information, and private communications. Therefore, one has to consider the security and privacy implication of unauthorized access to private information, e.g., due to loss or theft of a mobile device. To deal with this problem, continuous authentication (also known as active authentication) systems have been proposed. These systems continuously monitor users after the initial access to the mobile device in order to attest the users’ identity. This special session will present most recent active authentication research related to mobile and wearable devices, using various state-of-the‐art machine learning and computer vision techniques, including but not limited to behavioral and physiological biometrics‐based methods such as keystroke dynamics, behavior profiling, gait dynamics, and face-based methods. The session is expected to motivate the biometrics community for future research in this fast growing field.

Face Spoofing Detection in Real-World Applications

Organizers: Abdenour Hadid, University of Oulu, Finland; XiaoyiFeng, Northwestern Polytechnical University, China;Sebastien Marcel, IDIAP Research Institute, Switzerland

It is known that most of existing face recognition systems are vulnerable to spoofing attacks. A spoofing attack occurs when a person tries to bypass a face biometric system by presenting a fake face in front of the camera. The proposed anti-spoofing methods in the literature have shown very encouraging results on individual databases but lack generalization to varying nature of spoofing attacks that can be encountered in real-world applications. As the field evolves, new and more challenging databases are expected. The recent advances in machine learning (e.g. deep learning) are expected to play a key role in spoofing detection as well. This special session aims to bring together researchers working on face spoofing detection and related disciplines to present and discuss the recent developments in the field. This session will hopefully open a debate on new opportunities and new challenges in the area, and at unifying the efforts toward the development of new adequate tools, protocols and databases for evaluating and monitoring the progress in the field.

Ocular Biometrics in the Visible Spectrum

Organizers: Abhijit Das, Griffith University, Australia; Umapada Pal, Indian Statistical Institute, Kolkata, India; Michael Blumenstein, University Technology Sydney, Australia; Miguel A. Ferrer ULPGC, Spain

Ocular biometrics has received significant attention from researchers from industry, academia as well as government. The major reason behind the success of the ocular trait is due to the iris pattern. Although studies have reflected significant achievements in iris biometrics, its application in the visible spectrum (for mobile environments, at a distance/ in the wild) is an open challenge. Regardless of the satisfactory application of iris biometrics for lighter irises in the visible spectrum, its application for darker irises is an additional challenge. To address these challenges, two new traits namely conjunctival vasculature (sclera) and peri-ocular are proposed, that can be combine with iris to increase its biometric applicably in less constrain condition. However, the state-of-the-art related to recognition based on ocular biometrics in the visible spectrum still faces various open, unexplored, and unidentified challenges. Moreover, this problem is a fundamental problem in biometrics with broad economic (upcoming market growth of mobile biometrics will enjoy annual growth rate of over 150%, more demand of ocular biometrics applications at distance), social (secure mobile banking applications, secure surveillance, forensics applications) and scientific impact (development of detection, segmentation, characterization, identification algorithm in computer vision among others).

Therefore, this subject of research should receive structured and systematic multidisciplinary efforts in signal processing, pattern recognition, machine learning, and information fusion. Therefore, the goal of this special session is to bring together researchers and practitioners working in this area of biometrics to address a wide range of theoretical and practical issues related to systems based on this subject.

Representation Learning and Biometrics

Organizers: Mayank Vatsa, IIIT Delhi, India; Vishal Patel, Rutgers University, USA; Jiwen Lu, Tsinghua University, China Nasser M. Nasrabadi, West Virginia University, USA; Thirimachos Bourlai, West Virginia University, USA

Biometrics can be perceived as a research area with broader impact and success of large scale projects such as OBIM (previously known as US VISIT), India’s UID (Aadhaar), and FBI AFIS projects have touched the lives of billions. Biometrics is a multidisciplinary areas which includes sensor design, image processing, computer vision, pattern recognition, machine learning, and information fusion. Among these areas, advancements in pattern recognition and machine learning have helped biometrics research in addressing challenges such as identity recognition in unconstrained environment, very large scale identification, and feature learning from millions of data. If we focus on different steps of a biometrics pipeline, one of the important aspects is feature representation. While traditional approaches have focused on handcrafted features such as Local Binary Pattern and Scale Invariant Feature Transform, with the advent in computing technology learning representations have attracted several researchers worldwide.

As mentioned by Bengio et al. [1], the performance of machine learning methods is heavily dependent on the choice of data representation (or features) on which they are applied. Research in representation learning focuses on understanding the data distributions and utilize them in developing novel ways of extracting meaningful features that can help in improving the system performance. In biometrics, this improvement can be seen as higher accuracies and improved user convenience. There are multiple ingredients in representation learning such as deep learning, dictionary learning, non-convex optimization, and non-negative matrix factorization.

The objective of this special session on “Representation Learning and Biometrics” is to identify recent challenges in biometrics, present the algorithms following the representation learning approaches, and discuss the key results and future research directions. This special session will focus on all aspects of representation learning including deep learning and dictionary learning, with a particular emphasis on solving challenging biometrics problems and/or unsolved biometrics issues.

[1] Y. Bengio, A. Courville, P. Vincent, “Representation Learning: A Review and New Perspectives,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 8, pp. 1798-1828, Aug., 2013.