CITI Talk: “Deep Learning: history, models & challenges, with an application in signal processing and mobile authentification”, Ass. Prof. Christian Wolf (INSA Lyon, CITI-LIRIS), 22/02/2018, TD C

Leave a comment
christian

Title

Deep Learning: history, models & challenges, with an application in signal processing and mobile authentification

Abstract

Representation Learning (also known with its more popular title « Deep Learning ») consists in automatically learning layered and hierarchical representation with various layers abstraction from large amounts of data. This presentation will review the history of the field, the main actors and the major scientific challenges. We will first present a brief introduction into common deep models like convolutional neural networks and recurrent networks, before going more into detail of some selected applications in signal processing.

In particular, we present a large-scale study, exploring the capability of temporal deep neural networks in interpreting natural human kinematics and introduce the first method for active biometric authentication with mobile inertial sensors. This work has been done in collaboration with Google, where the first-of-its-kind dataset of human movements has been passively collected by 1500 volunteers using their smartphones daily over several months. We propose an optimized shift-invariant dense convolutional mechanism (DCWRNN) and incorporate the discriminatively-trained dynamic features in a probabilistic generative framework taking into account temporal characteristics. Our results demonstrate, that human kinematics convey important information about user identity and can serve as a valuable component of multi-modal authentication systems.

Biography

Christian WOLF is associate professor (Maitre de Conférences, HDR) at INSA de Lyon and LIRIS UMR 5205, a CNRS laboratory, since 2005. He is interested in computer vision and machine learning, deep learning, especially in the visual analysis of complex scenes in motion: gesture and activity recognition and pose estimation. In his work he puts an emphasis on models of complex interactions, on structured models, graphical models and on deep learning. He received his MSc in computer science from Vienna University of Technology (TU Wien) in 2000, and a PhD in computer science from INSA de Lyon, France, in 2003. In 2012 he obtained the habilitation diploma, also from INSA de Lyon. Since September 2017 Christian is on leave at INRIA, at the chroma work group at the CITI laboratory (“délégation INRIA”).


CITI Talk: “Hybrid High Performance Systems for Ultrascale Architectures”, Professor Carlos J. Barrios H. (UIS), 20/02/2018, 11h, TD C

Leave a comment
aaeaaqaaaaaaaadoaaaajgu3zddimge1lte4njmtngq3zs1hndu3ltuwnwixyme3mwm1zg

Abstract
Ultrascale architectures involve large-scale complex systems joining parallel and distributed computing infrastructures joining technology trends (i.e. RISC/ CISIC/ASICS systems) ultrascale software and hybrid models (i.e. cloud/Edge and Fog)  that will be extended in different escenarios. Precisely, this complexity propose special challenges from different points of view: sustainability, scalability, dynamicity, energy-aware, usability, data management, dependability  and more, in a software and hardware relationship. This presentation shows some towards about this ultrscale architectures, observing the different challenges and how it is possible to threat some of them, depending the context a point of view software/hardware and application context.

Bio
Professor Carlos J. Barrios H. is Doctor in Computer Science of the Nice–Côte d’Azur University in France. Researcher in High Performance Computing and Large Scale Architectures, he works in projects associated with hybrid/high performance architectures for science and ultrascale systems involving design, performance evaluation and implementation mechanisms. From 2012, Professor Barrios is the director of the High Performance and Scientific Computing Center of the Universidad Industrial de Santander in Colombia and assistant professor of the same university. At same time, he is the general chair of the Advanced Computing Services for Latin America and Caribbean and he’s involved in different HPC collaboration networks between Europe and Latin America. Contact: cbarrios@uis.edu.co


CITI Talk: “Clustering and Data Anonymization by Mutual Information”, Pablo Piantanida, Associate Professor at CentraleSupélec, TD D

Leave a comment
a_23916-filtered
Title
Clustering and Data Anonymization by Mutual Information

Abstract
In this talk, we first introduce  the Shannon theoretic
multi-clustering problem and investigate its properties, uncovering
connections with many other coding problems in the literature. The figure
of merit for this information-theoretic problem is mutual information, the
mathematical properties of which make the multi-clustering problem amenable
to techniques that could not be used in a general rate-distortion setting.
We start by considering the case of two sources, where we derive
singleletter bounds for the achievable region by connecting our setting to
hypothesis testing and pattern recognition recognition problems in the
information theory literature. We then generalize the problem setup to an
arbitrary number of sources and study a CEO problem with logarithmic loss
distortion and multiple description coding. Drawing from the theory of
submodular functions, we prove a tight inner and outer bound for the
resulting achievable region under a suitable conditional independence
assumption. Furthermore, we present a proof of the well-known two-function
case of a conjecture by Kumar and Courtade (2013), showing that the
dictator functions are essentially the only Boolean  functions maximizing
mutual information.  The key step in our proof is a careful analysis of the
Fourier spectrum of the two Boolean functions. Finally, we study
information-theoretic applications to the problem of statistical  data
anonymization via mutual information and deep learning methods in which the
identity of the data writer must remain private even from the learner.

Joint works with Dr. Georg Pichler (TU Wien, Austria), Prof. Gerald Matz
(TU Wien, Austria),  Clément Feutry (CentraleSupélec, France) and Yoshua
Bengio (Montréal, Canada)

Short biography
Pablo Piantanida received both B.Sc. in Electrical
Engineering and B.Sc. in Mathematics degrees from the University of Buenos
Aires (Argentina) in 2003, and the Ph.D. from Université Paris-Sud (Orsay,
France) in 2007. Since October 2007 he has joined the Laboratoire des
Signaux et Systèmes (L2S), at CentraleSupélec together with CNRS (UMR 8506)
and Université Paris-Sud, as an Associate Professor of Network Information
Theory. He is an IEEE Senior Member, coordinator of the Information Theory
and its Applications group (ITA) at L2S, and  coordinator of the
International Associate Laboratory (LIA) of the CNRS “Information, Learning
and Control” with several institutions in Montréal and General Co-Chair of
the 2019 IEEE International Symposium on Information Theory (ISIT). His
research interests lie broadly in information theory and its interactions
with other fields, including multi-terminal information and Shannon theory,
machine learning, statistical inference, communication mechanisms for
security and privacy, and representation learning.

CITI Talk: “Distributed Hypothesis Testing over Multi-User Channels”, Prof. Michele Wigger (Telecom ParisTech), 10am, amphitheater Émilie du Châtelet (Marie Curie Library-INSA de Lyon)

Leave a comment
me

Abstract

As part of the internet of things (IoT), the number of sensor nodes that wish to communicate with each other has exploded and is expected to further increase dramatically. Such an increase of communication devices inherently leads to involved communication and hypothesis testing scenarios, and thus calls for new coding and testing strategies. The talk presents new strategies and corresponding error exponents for different network scenarios, and it proves information-theoretic optimality of the proposed strategies in some cases. Special attention is given to scenarios where information collected at a sensor is desired at multiple decision centres and where communication is multi-hop involving sensor nodes as relays. In these networks, sensors generally compete for network resources, and relay sensors can process received information with sensed information or forward intermediate decisions to other nodes. Depending on the studied error exponents, some of these intermediate decisions require special protection mechanisms when sent over the network. The talk is based on joint work with Sadaf Salehkalaibar, Roy Timo, and Ligong Wang.


CITI Talk: “Coding for Cloud-RAN Downlink Channels”, Prof. Gerhard Kramer (TU Munchen), 9am, amphitheater Émilie du Châtelet (Marie Curie Library-INSA de Lyon)

Leave a comment
gerhard-kramer

Abstract

The downlink of a cloud radio accessnetwork (C-RAN) architecture can be modeled as a diamond network. The baseband unit (BBU) is connected to remote radio heads (RRHs) via fiber links that are modeled as rate-limited bit pipes. Bounds on the rates for reliable communication are evaluated for single-antenna RRHs. A lower bound is based on Marton’s coding, which facilitates dependence across the RRH signals. An upper bound uses Ozarow’s technique to augment the system with an auxiliary random variable. The bounds are studied over scalar Gaussian C-RANs and are shown to meet and characterize the capacity for interesting regimes of operation. The bounds are also evaluated for an abstract model: a noise-free binary adder channel (BAC). The capacity of the BAC is established for all ranges of bit-pipe capacities, which seems to yield a new combinatorial result on sum sets. This work is based on joint work with Shirin Saeedi Bidokhti and Shlomo Shamai.


CITI Talk: “Statistical Learning via Information Bottleneck”, Prof. Abdellatif Zaidi (Université Paris-Est Marne la Vallée), 4pm, room TD-E

Leave a comment
zaidi

Abstract

We connect the information flow in a neural network to sufficient statistics; and show how techniques that are rooted in information theory, such as the source-coding based information bottleneck method can lead to improved architectures, as well as a better understanding of the theoretical foundation of neural networks, viewed as a cascade compression network. We illustrate our results and view through some numerical examples.


PhD Defense: “Noisy Channel-Output Feedback in the Interference Channel” by Victor Quintero, 12th of December 2017, at 14h, in the amphitheater Émilie du Châtelet – Marie Curie Library – INSA de Lyon

Leave a comment
victor

Jury

  • Prof. Michèle WIGGER, Reviewer, Télécom Paristech, France.
  • Prof. Abdellatif ZAIDI, Reviewer, Université Paris-Est Marne la Vallée, France.
  • Prof. Inbar FIJALKOW, Examiner, Université de Cergy-Pontoise, France.
  • Prof. H. Vincent POOR, Examiner, Princeton University, USA.
  • Prof. Gerhard KRAMER, Examiner, Technische Universität München, Germany.
  • Prof. David GESBERT, Examiner, Eurecom, France.
  • Prof. Jean-Marie GORCE, Thesis Director, Université de Lyon, France.
  • Dr. Samir M. Perlaza, Thesis Advisor, INRIA, France.
  • Prof. Iñaki Esnaola, Guest, University of Sheffield, UK.

Abstract
In this thesis, the two-user Gaussian interference channel with noisy channel-output feedback (GIC-NOF) is studied from two perspectives: centralized and decentralized networks.

From the perspective of centralized networks, the fundamental limits of the two-user GIC- NOF are characterized by the capacity region. One of the main contributions of this thesis is an approximation to within a constant number of bits of the capacity region of the two-user GIC-NOF. This result is obtained thanks to the analysis of a simpler channel model, i.e., a two-user linear deterministic interference channel with noisy channel-output feedback (LDIC- NOF). The analysis to obtain the capacity region of the two-user LDIC-NOF provides the main insights required to analyze the two-user GIC-NOF.

From the perspective of decentralized networks, the fundamental limits of the two-user decentralized GIC-NOF (D-GIC-NOF) are characterized by the η-Nash equilibrium (η-NE) region. Another contribution of this thesis is an approximation to the η-NE region of the two-user GIC-NOF, with η > 1. As in the centralized case, the two-user decentralized LDIC-NOF (D-LDIC-NOF) is studied first and the lessons learnt are applied in the two-user D-GIC-NOF. The final contribution of this thesis consists of a closed-form answer to the question: “When does channel-output feedback enlarge the capacity or η-NE regions of the two-user GIC-NOF or two-user D-GIC-NOF?”. This answer is of the form: Implementing channel-output feedback in transmitter-receiver i enlarges the capacity or η-NE regions if the feedback SNR is beyond SNRi*, with i∈{1,2}. The approximate value of SNRi* is shown to be a function of all the other parameters of the two-user GIC-NOF or two-user D-GIC-NOF.


CITI Talk: “Body Impedance for Authentication, Key Generation and Device Pairing”, Kasper Rasmussen, Associate Professor (University of Oxford), December 8th 10:30, RobIoT room

Leave a comment
kasper_csox2

Abstract

Body Impedance is an effective biometric because each human body
exhibits a unique response to a signal applied at the palm of one hand
and measured in the palm of the other hand. We will see how body
impedance can be used both as a traditional biometric; to generate
cryptographic keys for use in traditional security purposes; or for
device pairing.

Our device pairing scheme is based on the idea that two devices can
pair, if they are physically held by the same person (at the same
time). To pair two devices, a person touches a conductive surface on
each device. While the person is in contact with both devices, the
human body acts as a transmission medium for intra-body communication
and the two devices can communicate through the body. This body
channel is used as part of a pairing protocol which allows the devices
to agree on a mutual secret and, at the same time, extract physical
features to verify that they are being held by the same person. We
prove that our device pairing protocol is secure with respect to a
strong threat model and we build a proof of concept set-up and conduct
experiments with 15 people to verify the idea in practice

Bio

Kasper Rasmussen is an Associate Professor in the Computer Science
Department at the University of Oxford. He joined the department in
2013 and in 2015 was awarded a University Research Fellowship from the
Royal Society in London. Prior to being at Oxford, Kasper Rasmussen
spent two years as a post-doc at University of California, Irvine.
Kasper Rasmussen did his Ph.D. with prof. Srdjan Capkun at the
Department of Computer Science at ETH Zurich (Switzerland), where he
worked on security issues relating to secure time synchronization and
secure localization with a particular focus on distance bounding. His
thesis won the “ETH Medal” for an outstanding dissertation from the
Swiss Federal Institute of Technology and he was additionally awarded
the Swiss National Science Foundation (SNSF), Fellowship for
prospective researchers.


PhD Defense: “Wi-Fi Tracking: Fingerprinting Attacks and Counter-Measures” by Célestin Matte, December 7th at 1:30pm in Amphi Chappe

Leave a comment
aaeaaqaaaaaaaadsaaaajgq1nmy4mte4lwe5zdutndm5nc1hnjgylwyymwnjnzvjnjkymw

Jury

Reviewers :
Nguyen, Benjamin, Professeur des universités, INSA Centre Val de Loire
Rasmussen, Kasper, Associate professor, University of Oxford

Members:
Chrisment, Isabelle, Professeur des universités, Université de Lorraine
Risset, Tanguy, Professeur des universités, INSA Lyon
Neumann, Christoph, Principal scientist, Technicolor

Supervisors :
Minier, Marine, Professeur des universités, Université de Lorraine
Cunche, Mathieu, Maître de conférences, Insa Lyon

Abstract

Wi-Fi Tracking: Fingerprinting Attacks and Counter-Measures

The recent spread of everyday-carried Wi-Fi-enabled devices (smartphones, tablets and wearable devices) comes with a privacy threat to their owner, and to society as a whole. These devices continuously emit signals which can be captured by a passive attacker using cheap hardware and basic knowledge. These signals contain a unique identifier,
called the MAC address. To mitigate the threat, device vendors are currently deploying a countermeasure on new devices: MAC address randomization. Unfortunately, we show that this mitigation, in its current state, is insufficient to prevent tracking.

To do so, we introduce several attacks, based on the content and the timing of emitted signals. In complement, we study implementations of MAC address randomization in some recent devices, and find a number of shortcomings limiting the efficiency of these implementations at preventing device tracking.

At the same time, we perform two real-world studies. The first one considers the development of actors exploiting this issue to install Wi-Fi tracking systems. We list some real-world installations and discuss their various aspects, including regulation, privacy implications, consent and public acceptance. The second one deals with the spread of MAC address randomization in the devices population.

Finally, we present two tools: an experimental Wi-Fi tracking system for testing and public awareness raising purpose, and a tool estimating the uniqueness of a device based on the content of its emitted signals even if the identier is randomized.