Publications

2021

• A Novel Pseudo-Bayesian Approach for Robust Multi-Ridge Detection and Mode Retrieval
  
  • Legros Quentin
  • Fourer Dominique
  , 2021, pp.1925--1929. This paper introduces a novel approach for extracting the elementary components present in an observed nonstationary mixture signal. Our technique based on a pseudo-Bayesian approach operates in the time-frequency plane and sequentially estimates the ridge of each component that is required for mode extraction. We compare our results with those obtained with the state-of-the-art Brevdo method which has shown its efficiency for disentangling multicomponent noisy signals. Our results reveal an improvement of the reconstruction performance when compared to the state of the art.
• Stack of Services for Context-Aware Systems: An Internet-Of-Things System Design Approach
  
  • Nguyen Quang-Duy
  • Roussey Catherine
  • Bellot Patrick
  • Chanet Jean-Pierre
  , 2021. The Internet of Things is an ideal world in which all computing devices from all over the world connect and exchange data through the Internet. This new scenario demands contextaware systems to evolve with new characteristics; thus, brings new challenges for system developers in system development. While addressing these challenges, this paper presents a system design approach based on a stack of 16 services specialized for context-aware systems. The approach enables system developers to focus more on services than hardware and software components. The case study of a smart irrigation context-aware system, also presented in this paper, is an example of using this design approach in practice. (10.1109/RIVF51545.2021.9642107)
  DOI : 10.1109/RIVF51545.2021.9642107
• Is This Still Normal? Putting Definitions of Timing Anomalies to the Test
  
  • Brandner Florian
  • Binder Benjamin
  • Asavoae Mihail
  • Hedia Belgacem Ben
  • Jan Mathieu
  , 2021, pp.139-148. (10.1109/RTCSA52859.2021.00024)
  DOI : 10.1109/RTCSA52859.2021.00024
• Structural Attack (and Repair) of Diffused-Input-Blocked-Output White-Box Cryptography
  
  • Carlet Claude
  • Guilley Sylvain
  • Mesnager Sihem
  IACR Transactions on Cryptographic Hardware and Embedded Systems, IACR, 2021, pp.57-87. In some practical enciphering frameworks, operational constraints may require that a secret key be embedded into the cryptographic algorithm. Such implementations are referred to as White-Box Cryptography (WBC). One technique consists of the algorithm’s tabulation specialized for its key, followed by obfuscating the resulting tables. The obfuscation consists of the application of invertible diffusion and confusion layers at the interface between tables so that the analysis of input/output does not provide exploitable information about the concealed key material.Several such protections have been proposed in the past and already cryptanalyzed thanks to a complete WBC scheme analysis. In this article, we study a particular pattern for local protection (which can be leveraged for robust WBC); we formalize it as DIBO (for Diffused-Input-Blocked-Output). This notion has been explored (albeit without having been nicknamed DIBO) in previous works. However, we notice that guidelines to adequately select the invertible diffusion ∅and the blocked bijections B were missing. Therefore, all choices for ∅ and B were assumed as suitable. Actually, we show that most configurations can be attacked, and we even give mathematical proof for the attack. The cryptanalysis tool is the number of zeros in a Walsh-Hadamard spectrum. This “spectral distinguisher” improves on top of the previously known one (Sasdrich, Moradi, Güneysu, at FSE 2016). However, we show that such an attack does not work always (even if it works most of the time).Therefore, on the defense side, we give a straightforward rationale for the WBC implementations to be secure against such spectral attacks: the random diffusion part ∅ shall be selected such that the rank of each restriction to bytes is full. In AES’s case, this seldom happens if ∅ is selected at random as a linear bijection of F322. Thus, specific care shall be taken. Notice that the entropy of the resulting ∅ (suitable for WBC against spectral attacks) is still sufficient to design acceptable WBC schemes. (10.46586/tches.v2021.i4.57-87)
  DOI : 10.46586/tches.v2021.i4.57-87
• Understanding the Within-Individual Variability of Forced Vital Capacity: An Exploitation of the NHANES III Spirometry Data
  
  • Guiard Yves
  , 2021. While there is an abundant literature on the distribution of spirometry statistics in various subsets of the human population, apparently little is known of the structure of the typically very small sample of measures that can be gathered during a spirometry session. This paper starts with a theoretical analysis of the relation linking the measure of forced vital capacity (FVC) to the parameter of total lungs capacity (TLC). Since the maximization effort exerted on FVC measures by the testees is opposed by the resistance of TLC, their impassable personal upper limit, a ceiling effect must take place on the continuum of FVC measurement. Two predictions follow concerning the within-subject distribution of FVC. One is that the distribution should be negatively skewed, the other is that its first and second moments should correlate negatively across sessions. These predictions were tested with the publicly available large-scale spirometry data collected by the Third National Health and Nutrition Examination Survey. Using original data processing techniques especially devised to unveil the shape of small session samples of FVC measures, the paper reports highly consistent confirmatory evidence, based on the analysis of thousands of individual test sessions, that a typical session sample of FVC is indeed strongly skewed negatively and that the session mean and the session standard deviation of FVC do indeed bear a strong negative correlation. Several implications of these results are discussed, some of which cut across the frontiers of respirology. It is suggested that the procedural rigor and simplicity of spirometry testing make it a privileged paradigm for understanding quantitative performance measurement in general.
• Chaos synchronization in mid-infrared quantum cascade lasers for private free-space communication
  
  • Spitz Olivier
  • Herdt Andreas
  • Maisons Gregory
  • Carras Mathieu
  • Elsaser Wolfgang
  • Grillot Frederic
  , 2021, pp.1-2. (10.1109/RAPID51799.2021.9521424)
  DOI : 10.1109/RAPID51799.2021.9521424
• A Novel Estimator of Mutual Information for Learning to Disentangle Textual Representations
  
  • Colombo Pierre
  • Piantanida Pablo
  • Clavel Chloé
  , 2021. Learning disentangled representations of textual data is essential for many natural language tasks such as fair classification, style transfer and sentence generation, among others. The existent dominant approaches in the context of text data either rely on training an adversary (discriminator) that aims at making attribute values difficult to be inferred from the latent code or rely on minimising variational bounds of the mutual information between latent code and the value attribute. However, the available methods suffer of the impossibility to provide a fine-grained control of the degree (or force) of disentanglement. In contrast to adversarial methods, which are remarkably simple, although the adversary seems to be performing perfectly well during the training phase, after it is completed a fair amount of information about the undesired attribute still remains. This paper introduces a novel variational upper bound to the mutual information between an attribute and the latent code of an encoder. Our bound aims at controlling the approximation error via the Renyi's divergence, leading to both better disentangled representations and in particular, a precise control of the desirable degree of disentanglement than state-of-the-art methods proposed for textual data. Furthermore, it does not suffer from the degeneracy of other losses in multi-class scenarios. We show the superiority of this method on fair classification and on textual style transfer tasks. Additionally, we provide new insights illustrating various trade-offs in style transfer when attempting to learn disentangled representations and quality of the generated sentence. (10.48448/dqvn-s462)
  DOI : 10.48448/dqvn-s462
• Constrained projective dynamics: real-time simulation of deformable objects with energy-momentum conservation
  
  • Kee Min Hyung
  • Um Kiwon
  • Jeong Wooseok
  • Han Junghyun
  ACM Transactions on Graphics, Association for Computing Machinery, 2021, 40 (4), pp.1-12. This paper proposes a novel energy-momentum conserving integration method. Adopting Projective Dynamics, the proposed method extends its unconstrained minimization for time integration into the constrained form with the position-based energy-momentum constraints. This resolves the well-known problem of unwanted dissipation of energy and momenta without compromising the real-time performance and simulation stability. The proposed method also enables users to directly control the energy and momenta so as to easily create the vivid deformable and global motions they want, which is a fascinating feature for many real-time applications such as virtual/augmented reality and games. (10.1145/3450626.3459878)
  DOI : 10.1145/3450626.3459878
• Spectral estimation for non-linear long range dependent discrete time trawl processes
  
  • Doukhan Paul
  • Roueff François
  • Rynkiewicz Joseph
  Electronic Journal of Statistics, Shaker Heights, OH : Institute of Mathematical Statistics, 2021, 14 (2), pp.3157 - 3191. Discrete time trawl processes constitute a large class of time series parameterized by a trawl sequence (a j) j∈N and defined though a sequence of independent and identically distributed (i.i.d.) copies of a continuous time process (γ(t)) t∈R called the seed process. They provide a general framework for modeling linear or non-linear long range dependent time series. We investigate the spectral estimation, either pointwise or broadband, of long range dependent discrete-time trawl processes. The difficulty arising from the variety of seed processes and of trawl sequences is twofold. First, the spectral density may take different forms, often including smooth additive correction terms. Second, trawl processes with similar spectral densities may exhibit very different statistical behaviors. We prove the consistency of our estimators under very general conditions and we show that a wide class of trawl processes satisfy them. This is done in particular by introducing a weighted weak dependence index that can be of independent interest. The broadband spectral estimator includes an estimator of the long memory parameter. We complete this work with numerical experiments to evaluate the finite sample size performance of this estimator for various integer valued discrete time trawl processes. (10.1214/20-EJS1742)
  DOI : 10.1214/20-EJS1742
• Meta-learning for Classifying Previously Unseen Data Source into Previously Unseen Emotional Categories
  
  • Guibon Gaël
  • Labeau Matthieu
  • Flamein Hélène
  • Lefeuvre Luce
  • Clavel Chloé
  , 2021. In this paper, we place ourselves in a classification scenario in which the target classes and data type are not accessible during training. We use a meta-learning approach to determine whether or not meta-trained information from common social network data with fine-grained emotion labels can achieve competitive performance on messages labeled with different emotion categories. We leverage fewshot learning to match with the classification scenario and consider metric learning based meta-learning by setting up Prototypical Networks with a Transformer encoder, trained in an episodic fashion. This approach proves to be effective for capturing meta-information from a source emotional tag set to predict previously unseen emotional tags. Even though shifting the data type triggers an expected performance drop, our meta-learning approach achieves decent results when compared to the fully supervised one.
• DAG amendment for inverse control of parametric shapes
  
  • Michel Elie
  • Boubekeur Tamy
  ACM Transactions on Graphics, Association for Computing Machinery, 2021, 40 (4), pp.1 - 14. (10.1145/3450626.3459823)
  DOI : 10.1145/3450626.3459823
• Stochastic optimization with momentum: convergence, fluctuations, and traps avoidance
  
  • Barakat Anas
  • Bianchi Pascal
  • Hachem Walid
  • Schechtman Sholom
  Electronic Journal of Statistics, Shaker Heights, OH : Institute of Mathematical Statistics, 2021, 15 (2), pp.3892-3947. In this paper, a general stochastic optimization procedure is studied, unifying several variants of the stochastic gradient descent such as, among others, the stochastic heavy ball method, the Stochastic Nesterov Accelerated Gradient algorithm (S-NAG), and the widely used Adam algorithm. The algorithm is seen as a noisy Euler discretization of a nonautonomous ordinary differential equation, recently introduced by Belotto da Silva and Gazeau, which is analyzed in depth. Assuming that the objective function is non-convex and differentiable, the stability and the almost sure convergence of the iterates to the set of critical points are established. A noteworthy special case is the convergence proof of SNAG in a nonconvex setting. Under some assumptions, the convergence rate is provided under the form of a Central Limit Theorem. Finally, the non-convergence of the algorithm to undesired critical points, such as local maxima or saddle points, is established. Here, the main ingredient is a new avoidance of traps result for non-autonomous settings, which is of independent interest. (10.1214/21-EJS1880)
  DOI : 10.1214/21-EJS1880
• Lightweight blockchain processing: case study: scanned document tracking on Tezos blockchain
  
  • Allouche Mohamed
  • Frikha Tarek
  • Mitrea Mihai
  • Memmi Gérard
  • Chaabane Faten
  Applied Sciences, Multidisciplinary digital publishing institute (MDPI), 2021, 11 (15), pp.7169:1-7169:17. To bridge the current gap between the Blockchain expectancies and their intensive computation constraints, the present paper advances a lightweight processing solution, based on a load-balancing architecture, compatible with the lightweight/embedding processing paradigms. In this way, the execution of complex operations is securely delegated to an off-chain general-purpose computing machine while the intimate Blockchain operations are kept on-chain. The illustrations correspond to an on-chain Tezos configuration and to a multiprocessor ARM embedded platform (integrated into a Raspberry Pi). The performances are assessed in terms of security, execution time, and CPU consumption when achieving a visual document fingerprint task. It is thus demonstrated that the advanced solution makes it possible for a computing intensive application to be deployed under severely constrained computation and memory resources, as set by a Raspberry Pi 3. The experimental results show that up to nine Tezos nodes can be deployed on a single Raspberry Pi 3 and that the limitation is not derived from the memory but from the computation resources. The execution time with a limited number of fingerprints is 40% higher than using a classical PC solution (value computed with 95% relative error lower than 5%). (10.3390/app11157169)
  DOI : 10.3390/app11157169
• Improving Reproducibility in Machine Learning Research (A Report from the NeurIPS 2019 Reproducibility Program)
  
  • Pineau Joelle
  • Vincent-Lamarre Philippe
  • Sinha Koustuv
  • Larivière Vincent
  • Beygelzimer Alina
  • d'Alché-Buc Florence
  • Fox Emily
  • Larochelle Hugo
  Journal of Machine Learning Research, Microtome Publishing, 2021, 22. One of the challenges in machine learning research is to ensure that presented and published results are sound and reliable. Reproducibility, that is obtaining similar results as presented in a paper or talk, using the same code and data (when available), is a necessary step to verify the reliability of research findings. Reproducibility is also an important step to promote open and accessible research, thereby allowing the scientific community to quickly integrate new findings and convert ideas to practice. Reproducibility also promotes the use of robust experimental workflows, which potentially reduce unintentional errors. In 2019, the Neural Information Processing Systems (NeurIPS) conference, the premier international conference for research in machine learning, introduced a reproducibility program, designed to improve the standards across the community for how we conduct, communicate, and evaluate machine learning research. The program contained three components: a code submission policy, a community-wide reproducibility challenge, and the inclusion of the
• Asymptotically-Good Arithmetic Secret Sharing over Z/pℓZ with Strong Multiplication and Its Applications to Efficient MPC
  
  • Cramer Ronald
  • Rambaud Matthieu
  • Xing Chaoping
  , 2021, 12827, pp.656-686. (10.1007/978-3-030-84252-9_22)
  DOI : 10.1007/978-3-030-84252-9_22
• Reflective Electroabsorption Modulators for Beyond 25 Gb/s Colorless Transmissions
  
  • Atra Kebede
  • Cerulo Giancarlo
  • Provost Jean-Guy
  • Mekhazni Karim
  • Calo Cosimo
  • Pommereau Frederic
  • Gomez Carmen
  • Wilk Arnaud
  • Blache Fabrice
  • Fortin Catherine
  • Decobert Jean
  • Martin Florence F.
  • Derouin Estelle
  • Caillaud Christophe
  • Ware Cédric
  • Erasme Didier
  • Mallecot Franck
  • Achouche Mohand
  Journal of Lightwave Technology, Institute of Electrical and Electronics Engineers (IEEE)/Optical Society of America(OSA), 2021, 39 (15), pp.5035-5041. We present a complete characterization of reflective electroabsorption modulators (EAMs) monolithically integrated with semiconductor optical amplifiers (SOAs), components that are capable of operating beyond 50 Gb/s in the C-band. The devices are based on GaInAsP multiple quantum wells on InP substrate, leveraging semi-insulating buried heterostructure waveguide definition and butt-joint integration technologies. Different device configurations, based on 80-μm and 150-μm long EAMs, are fabricated and characterized in both static and dynamic modes. The frequency response of the 80-μm long EAM is still flat at 26.5 GHz (setup upper limit) whereas the 150-μm long EAM exhibits a 3-dB cutoff bandwidth of 23 GHz. A zero-chirp is achieved for EAM reverse bias voltages between −1.2 V and −1.5 V depending on the wavelength. Under large-signal modulation, the frequency chirp induced by the shorter EAM is almost half that of the longer EAM, with their respective peak values being +1.5/−2 GHz and +3.2/−3.7 GHz (rising/falling edges) at 1545 nm (−1.3 V bias, 2.6 V voltage swing). We obtained high dynamic extinction ratios of ~14.5 dB and ~8 dB from the longer and the shorter EAMs, respectively, when they are operated at 25 Gb/s using non-return-to-zero coding. Finally, we achieved 12 km and 16 km colorless transmissions in the C-band (between 1530 nm and 1545 nm) over a standard single-mode fiber without equalization using the 150-μm and the 80-μm EAMs, respectively, with 4.5 dB and 2.5 dB dispersion penalties at a bit error rate of 10E−3. (10.1109/JLT.2021.3079987)
  DOI : 10.1109/JLT.2021.3079987
• This is IT: A Primer on Shannon’s Entropy and Information
  
  • Rioul Olivier
  , 2021, 78. What is Shannon’s information theory (IT)? Despite its continued impact on our digital society, Claude Shannon’s life and work is still unknown to numerous people. In this tutorial, we review many aspects of the concept of entropy and information from a historical and mathematical point of view. The text is structured into small, mostly independent sections, each covering a particular topic. For simplicity we restrict our attention to one-dimensional variables and use logarithm and exponential notations log and exp without specifying the base. We culminate with a simple exposition of a recent proof (2017) of the entropy power inequality (EPI), one of the most fascinating inequalities in the theory.
• Assessing adversarial training effect on IDSs and GANs
  
  • Chaitou Hassan
  • Robert Thomas
  • Leneutre Jean
  • Pautet Laurent
  , 2021, pp.543-550. Deep neural network-based Intrusion Detection Systems (IDSs) are gaining popularity to improve anomaly detection accuracy and robustness. Yet, Deep neural network (DNN) models have been shown to be vulnerable to adversarial attacks. An attacker can use a generator, here a Generative Adversarial Network, to alter an attack so that the IDS model misclassify it as normal network traffic. There is a race between adversarial attacks and mechanisms to make robust IDSs, like Adversarial Training. To our knowledge, no study thoroughly assesses how attack generators or IDS training is sensitive to parameters controlling resources spent during training. Such results provide insights on how much to spend on IDS training. This paper presents the outcome of this assessment for GANs vs adversarial training. Interestingly, it shows that GANs' evasion capabilities are either very good or poor, with almost no average cases. Resources impact the likelihood of obtaining an efficient generator. (10.1109/CSR51186.2021.9527949)
  DOI : 10.1109/CSR51186.2021.9527949
• Study of minimum mean square error optimal equalizers for 50Gbit/s high speed passive optical networks
  
  • Nogueira Sampaio Flávio Andre
  • Genay Naveena
  • Pincemin Erwan
  • Le Bidan Raphaël
  • Jaouën Yves
  • Bourgart Fabrice
  , 2021. We evaluated here the positive impact of MMSE equalizers on 50Gbit/s HS-PONs. Thanks to numerical simulations, we show that MMSE-LE and DFE allows to reach appreciable optical budget of 29dB for pre-FEC BER of 10–2. (10.1364/SPPCOM.2021.SpF1E.2)
  DOI : 10.1364/SPPCOM.2021.SpF1E.2
• Low Complexity Convolutional Neural Networks for Equalization in Optical Fiber Transmission
  
  • Abu-Romoh Mohannad
  • Costa Nelson
  • Napoli Antonio
  • Pedro João
  • Jaouën Yves
  • Yousefi Mansoor
  , 2022. A convolutional neural network is proposed to mitigate fiber transmission effects, achieving a five-fold reduction in trainable parameters compared to alternative equalizers, and 3.5 dB improvement in MSE compared to DBP with comparable complexity.
• Multiple and Reproducible Fault Models on Micro-Controller using Electromagnetic Fault Injection
  
  • Khuat Vanthanh
  • Trabelsi Oualid
  • Sauvage Laurent
  • Danger Jean-Luc
  , 2021. (10.1109/EMC/SI/PI/EMCEurope52599.2021.9559288)
  DOI : 10.1109/EMC/SI/PI/EMCEurope52599.2021.9559288
• IETF Draft, "Identity Module for TLS Version 1.3" https://datatracker.ietf.org/doc/html/draft-urien-tls-im-05
  
  • Urien Pascal
  , 2021.
• Brief Announcement: Malicious Security Comes for Free in Consensus with Leaders
  
  • Abspoel Mark
  • Attema Thomas
  • Rambaud Matthieu
  , 2021, pp.195-198. (10.1145/3465084.3467953)
  DOI : 10.1145/3465084.3467953
• Procédé d’entraînement d’un modèle statistique pour qu’il soit configuré pour être utilisé pour recommander, à partir d’un média d’un premier type, un média d’un deuxième type, et système associé
  
  • Prétet Laure
  • Richard Gaël
  • Peeters Geoffroy
  • Souchier Clément
  , 2021.
• POSTER: Resistance Analysis of Two AES-Like Against the Boomerang Attack
  
  • Debesse Laetitia
  • Mesnager Sihem
  • Msahli Mounira
  , 2021, 12809, pp.485-489. (10.1007/978-3-030-81645-2_27)
  DOI : 10.1007/978-3-030-81645-2_27