Publications

2020

• Technologies quantiques et cryptographie post-quantique : vision prospective et positionnement français
  
  • Alleaume Romain
  , 2020.
• Surrogate modeling of indoor down-link human exposure based on sparse polynomial chaos expansion
  
  • Liu Zicheng
  • Lesselier Dominique
  • Sudret Bruno
  • Wiart Joe
  International Journal for Uncertainty Quantification, Begell House Publishers, 2020, 10 (2), pp.145-163. Human exposure induced by wireless communication systems increasingly draws the public attention. Here, an indoordown-link scenario is concerned and the exposure level is statistically analyzed. The electromagnetic field (EMF)emitted by a WiFi box is measured and electromagnetic dosimetry features are evaluated from the whole-body specificabsorption rate as computed with a Finite-Difference Time-Domain (a.k.a. FDTD) code. Due to computational cost, astatistical analysis is performed based on a surrogate model, which is constructed by means of so-called sparse polynomialchaos expansion (PCE), where the inner cross validation (ICV) is used to select the optimal hyperparametersduring the model construction and assess the model performance. However, the ICV error is optimized and the modelassessment tends to be overly optimistic with small experimental configurations. The method of cross-model validation is usedand outer cross validation is carried out for the model assessment. The effects of the data preprocessing are investigatedas well. Based on the surrogate model, the global sensitivity of the exposure to input parameters is analyzed from Sobol’indices. (10.1615/Int.J.UncertaintyQuantification.2020031452)
  DOI : 10.1615/Int.J.UncertaintyQuantification.2020031452
• Joint Subcarrier and Power Allocation in NOMA: Optimal and Approximate Algorithms
  
  • Salaun Lou
  • Coupechoux Marceau
  • Chen Chung Shue
  IEEE Transactions on Signal Processing, Institute of Electrical and Electronics Engineers, 2020, 68 (1), pp.2215 - 2230. Non-orthogonal multiple access (NOMA) is a promising technology to increase the spectral efficiency and enable massive connectivity in 5G and future wireless networks. In contrast to orthogonal schemes, such as OFDMA, NOMA multiplexes several users on the same frequency and time resource. Joint subcarrier and power allocation problems (JSPA) in NOMA are NP-hard to solve in general. In this family of problems, we consider the weighted sum-rate (WSR) objective function as it can achieve various tradeoffs between sum-rate performance and user fairness. Because of JSPA's intractability, a common approach in the literature is to solve separately the power control and subcarrier allocation (also known as user selection) problems, therefore achieving sub-optimal result. In this work, we first improve the computational complexity of existing single-carrier power control and user selection schemes. These improved procedures are then used as basic building blocks to design new algorithms, namely OPT-JSPA, ε-JSPA and GRAD-JSPA. OPT-JSPA computes an optimal solution with lower complexity than current optimal schemes in the literature. It can be used as a benchmark for optimal WSR performance in simulations. However, its pseudo-polynomial time complexity remains impractical for real-world systems with low latency requirements. To further reduce the complexity, we propose a fully polynomial-time approximation scheme called ε-JSPA. Since, no approximation has been studied in the literature, ε-JSPA stands out by allowing to control a tight trade-off between performance guarantee and complexity. Finally, GRAD-JSPA is a heuristic based on gradient descent. Numerical results show that it achieves near-optimal WSR with much lower complexity than existing optimal methods. (10.1109/TSP.2020.2982786)
  DOI : 10.1109/TSP.2020.2982786
• Flexible and Context-Specific AI Explainability: A Multidisciplinary Approach
  
  • Beaudouin Valérie
  • Bloch Isabelle
  • Bounie David
  • Clémençon Stéphan
  • d'Alché-Buc Florence
  • Eagan James R
  • Maxwell Winston
  • Mozharovskyi Pavlo
  • Parekh Jayneel
  SSRN Electronic Journal, Elsevier, 2020, pp.1-66. The recent enthusiasm for artificial intelligence (AI) is due principally to advances in deep learning. Deep learning methods are remarkably accurate, but also opaque, which limits their potential use in safety-critical applications. To achieve trust and accountability, designers and operators of machine learning algorithms must be able to explain the inner workings, the results and the causes of failures of algorithms to users, regulators, and citizens. The originality of this paper is to combine technical, legal and economic aspects of explainability to develop a framework for defining the "right" level of explain-ability in a given context. We propose three logical steps: First, define the main contextual factors, such as who the audience of the explanation is, the operational context, the level of harm that the system could cause, and the legal/regulatory framework. This step will help characterize the operational and legal needs for explanation, and the corresponding social benefits. Second, examine the technical tools available, including post hoc approaches (input perturbation, saliency maps...) and hybrid AI approaches. Third, as function of the first two steps, choose the right levels of global and local explanation outputs, taking into the account the costs involved. We identify seven kinds of costs and emphasize that explanations are socially useful only when total social benefits exceed costs. (10.2139/ssrn.3559477)
  DOI : 10.2139/ssrn.3559477
• Advances in security research in the Asiacrypt region
  
  • Phan Raphaël Cw
  • Abe Masayuki
  • Batten Lynn
  • Cheon Jung Hee
  • Dawson Ed
  • Galbraith Steven
  • Guo Jian
  • Hui Lucas
  • Kim Kwangjo
  • Lai Xuejia
  • Lee Dong Hoon
  • Matsui Mitsuru
  • Matsumoto Tsutomu
  • Moriai Shiho
  • Nguyen Phong
  • Pei Dingyi
  • Phan Duong Hieu
  • Pieprzyk Josef
  • Wang Huaxiong
  • Wolfe Hank
  • Wong Duncan
  • Wu Tzong-Chen
  • Yang Bo-Yin
  • Yiu Siu-Ming
  • Yu Yu
  • Zhou Jianying
  Communications of the ACM, Association for Computing Machinery, 2020, 63 (4), pp.76-81. Members of the International Association for Cryptologic Research explore regional work and collaboration activities. (10.1145/3378428)
  DOI : 10.1145/3378428
• Design of a combinatorial double auction for local energy markets
  
  • Kiedanski Diego
  • Kofman Daniel
  • Orda Ariel
  , 2020. Local energy markets allow neighbours to exchange energy among them. Their traditional implementation using sequential auctions has proven to be inefficient and even counterproductive in some cases. In this paper we propose a combinatorial double auction for the exchange of energy for several time-slots simultaneously. We suppose that participants have a flexible demand; flexibility being obtained, for example, by the usage of a battery. We show the benefits of the approach and we provide an example of how it can improve the utility of all the participants in the market.
• Smart dipole arrays for radio channel enhancement
  
  • Garcia Juan Carlos Bucheli
  • Sibille Alain
  • Kamoun Mohamed
  , 2020.
• Wide frequency characterization of Intra-Body Communication for Leadless Pacemakers
  
  • Maldari Mirko
  • Albatat Mohammad
  • Bergsland Jacob
  • Haddab Youcef
  • Jabbour Chadi
  • Desgreys Patricia
  IEEE Transactions on Biomedical Engineering, Institute of Electrical and Electronics Engineers, 2020, pp.1-1. (10.1109/TBME.2020.2980205)
  DOI : 10.1109/TBME.2020.2980205
• Resource allocation and optimization for the non-orthogonal multiple access
  
  • Salaün Lou
  , 2020. Non-orthogonal multiple access (NOMA) is a promising technology to increase the spectral efficiency and enable massive connectivity in future wireless networks. In contrast to orthogonal schemes, such as OFDMA, NOMA can serve multiple users on the same frequency and time resource by superposing their signal in the power domain. One of the key challenges for radio resource management (RRM) in NOMA systems is to solve the joint subcarrier and power allocation (JSPA) problem. In this thesis, we present a novel optimization framework to study a general class of JSPA problems. This framework employs a generic objective function which can be used to represent the popular weighted sum-rate (WSR), proportional fairness, harmonic mean and max-min fairness utilities. Our work also integrates various realistic constraints. We prove under this framework that JSPA is NP-hard to solve in general. In addition, we study its computational complexity and approximability in various special cases, for different objective functions and constraints. In this framework, we first consider the WSR maximization problem subject to cellular power constraint. We propose three new algorithms: Opt-JSPA computes an optimal solution with lower complexity than current optimal schemes in the literature. It can be used as an optimal benchmark in simulations. However, its pseudo-polynomial time complexity remains impractical for real-world systems with low latency requirements. To further reduce the complexity, we propose a fully polynomial-time approximation scheme called Ɛ-JSPA, which allows tight trade-offs between performance guarantee and complexity. To the best of our knowledge, Ɛ-JSPA is the first polynomial-time approximation scheme proposed for this problem. Finally, Grad-JSPA is a heuristic based on gradient descent. Numerical results show that it achieves near-optimal WSR with much lower complexity than existing optimal methods. As a second application of our framework, we study individual power constraints. Power control is solved optimally by gradient descent methods. Then, we develop three heuristics: DGA, DPGA and DIWA, which solve the JSPA problem for centralized and distributed settings. Their performance and computational complexity are compared through simulations.
• Wireless communication in presence of digitally controllable scatterers: channel decomposition and capacity analysis
  
  • Garcia Juan Carlos Bucheli
  • Kamoun Mohamed
  • Sibille Alain
  , 2020.
• Geometric operators for 3D modeling using dictionary-based shape representations
  
  • Lescoat Thibault
  , 2020. In this thesis, we study high-level 3D shape representations and developed the algorithm primitives necessary to manipulate shapes represented as a composition of several parts. We first review existing representations, starting with the usual low-level ones and then expanding on a high-level family of shape representations, based on dictionaries. Notably, we focus on representing shapes via a discrete composition of atoms from a dictionary of parts.We observe that there was no method to smoothly blend non-overlapping atoms while still looking plausible. Indeed, most methods either required overlapping parts or do not preserve large-scale details. Moreover, very few methods guaranteed the exact preservation of the input, which is very important when dealing with artist-authored meshes to avoid destroying the artist's work. We address this challenge by proposing a composition operator that is guaranteed to exactly keep the input while also propagating large-scale details.To improve the speed of our composition operator and allow interactive edition, we propose to simplify the input parts prior to completing them. This allow us to interactively previsualize the composition of large meshes. For this, we introduce a method to simplify a detailed mesh to a coarse one by preserving the large details. While more constrained than related approaches that do not produce a mesh, our method still yields faithful outputs.
• Effect of Aging on PUF Modeling Attacks based on Power Side-Channel Observations
  
  • Danger Jean-Luc
  • Cheng Wei
  • Guilley Sylvain
  • Karimi Naghmeh
  • Kroeger Trevor
  , 2020, pp.454-459. (10.23919/DATE48585.2020.9116428)
  DOI : 10.23919/DATE48585.2020.9116428
• O-band Reflective Electroabsorption Modulator for 50 Gb/s NRZ and PAM-4 Colorless Transmission
  
  • Atra Kebede
  • Cerulo Giancarlo
  • Provost Jean-Guy
  • Jorge Filipe
  • Blache Fabrice
  • Mekhazni Karim
  • Garreau Alexandre
  • Pommereau Frederic
  • Gomez Carmen
  • Fortin Catherine
  • Ware Cédric
  • Erasme Didier
  • Mallecot Franck
  • Achouche Mohand
  , 2020 (M2B.1). (10.1364/OFC.2020.M2B.1)
  DOI : 10.1364/OFC.2020.M2B.1
• Effects of dynamics and optical feedback on hybrid III-V/Si semiconductor lasers
  
  • Cadavid Gomez Sandra
  , 2020. Photonic Integrated Circuits (PIC) have become key elements to perform broadband transmission and reception functions in optical communication networks. This thesis provides information on hybrid semiconductor lasers (SCL) consisting of an active layer of III-V materials on a silicon-on-insulator (SOI) substrate to jointly explode the emitting properties of III-V layers and the numerous advantages offered by Si for on-chip applications. Due to the significant technological developments in electronics, this hybrid approach is well positioned to meet the high volume requirements for short distance transmission and access networks at a lower cost. However, several challengest still exist such as the lack of effective light sources and isolator-free devices. From a monolithic perspective of a PIC hetereogeneous integration, it is essential to ensure that the parasitic reflections that may stem from multiple locations do not affect the stability of the laser. Specifically, III-V hybrid components on Si appear to have many potential sources of reflections that can create centimeter external cavities in addition to those naturally produced inside the fiber in the order of several meters. Therefore, the work presented herein aims at understanding the behavior of III-V/Si SCLs when subjected to a variation of optical feedback, explores the basics aspects of chaotic dynamics, and investigates potential applications suitable for optical telecommunication systems in an attempt to meet the existing an emerging high speed requirements.
• NeuroQuery, comprehensive meta-analysis of human brain mapping
  
  • Dockès Jérôme
  • Poldrack Russell
  • Primet Romain
  • Gözükan Hande
  • Yarkoni Tal
  • Suchanek Fabian M
  • Thirion Bertrand
  • Varoquaux Gaël
  eLife, eLife Sciences Publication, 2020. Reaching a global view of brain organization requires assembling evidence on widely different mental processes and mechanisms. The variety of human neuroscience concepts and terminology poses a fundamental challenge to relating brain imaging results across the scientific literature. Existing meta-analysis methods perform statistical tests on sets of publications associated with a particular concept. Thus, large-scale meta-analyses only tackle single terms that occur frequently. We propose a new paradigm, focusing on prediction rather than inference. Our multivariate model predicts the spatial distribution of neurological observations, given text describing an experiment, cognitive process, or disease. This approach handles text of arbitrary length and terms that are too rare for standard meta-analysis. We capture the relationships and neural correlates of 7 547 neuroscience terms across 13 459 neuroimaging publications. The resulting meta-analytic tool, neuroquery.org, can ground hypothesis generation and data-analysis priors on a comprehensive view of published findings on the brain.
• Attention-based Fusion for Multi-source Human Image Generation
  
  • Lathuilière Stéphane
  • Sangineto Enver
  • Siarohin Aliaksandr
  • Sebe Nicu
  , 2019. We present a generalization of the person-image generation task, in which a human image is generated conditioned on a target pose and a set X of source appearance images. In this way, we can exploit multiple, possibly complementary images of the same person which are usually available at training and at testing time. The solution we propose is mainly based on a local attention mechanism which selects relevant information from different source image regions, avoiding the necessity to build specific generators for each specific cardinality of X. The empirical evaluation of our method shows the practical interest of addressing the person-image generation problem in a multi-source setting.
• On the Mean Interference-to-Signal Ratio in Spatially Correlated Cellular Networks
  
  • Wang Shanshan
  • Di Renzo Marco
  IEEE Wireless Communications Letters, IEEE comsoc, 2020, 9 (3), pp.358-362. The Inhomogeneous Double Thinning (IDT) approach is a tractable and general approximation for analyzing cellular networks in which the locations of the base stations are spatially correlated. Usually, however, the parameters of the approximation depend on the intensity and the specific parameters of the point process. Therefore, it is often difficult to unveil the impact of the system parameters from the resulting analytical framework. In this letter, we focus our attention on spatially repulsive cellular networks, and introduce a new parameterization for the IDT approach that is suitable for analysis. By assuming a bounded path-loss model, we prove the following trends for the Mean Interference-to-Signal Ratio (MISR): (i) the MISR monotonically decreases as the path-loss exponent increases and the deployment density of the base stations decreases; and (ii) the difference between the MISRs of Poisson and non-Poisson cellular networks monotonically decreases as the path-loss exponent increases and the density of the base stations decreases. As case studies, we specialize the proposed parametrization to the β-Ginibre and square lattice point processes. We prove, in particular, that the MISR monotonically decreases as β increases. These findings are shown to be in agreement with Monte Carlo simulations, and with numerical and analytical studies reported in prior works, thus substantiating the validity of the proposed parametrization. (10.1109/lwc.2019.2955084)
  DOI : 10.1109/lwc.2019.2955084
• Equivalent Rewritings on Path Views with Binding Patterns
  
  • Romero Julien
  • Preda Nicoleta
  • Amarilli Antoine
  • Suchanek Fabian M.
  , 2020, 12123 LNCS, pp.446-462. A view with a binding pattern is a parameterized query on a database. Such views are used, e.g., to model Web services. To answer a query on such views, the views have to be orchestrated together in execution plans. We show how queries can be rewritten into equivalent execution plans, which are guaranteed to deliver the same results as the query on all databases. We provide a correct and complete algorithm to find these plans for path views and atomic queries. Finally, we show that our method can be used to answer queries on real-world Web services. (10.1007/978-3-030-49461-2_26)
  DOI : 10.1007/978-3-030-49461-2_26
• CardBot: Towards an affordable humanoid robot platform for Wizard of Oz Studies in HRI
  
  • Krishna Sooraj
  • Pelachaud Catherine
  , 2020, pp.73-73. CardBot is a cardboard based programmable humanoid robot platform designed for inexpensive and rapid prototyping of Wizard of Oz interactions in HRI incorporating technologies such as Arduino, Android and Unity3d. The table demonstration showcases the design of the CardBot and its wizard controls such as animating the movements, coordinating speech and gaze etc for orchestrating an interaction. (10.1145/3371382.3378203)
  DOI : 10.1145/3371382.3378203
• Technology Trends for Mixed QKD/WDM Transmission up to 80 km
  
  • Alleaume Romain
  • Aymeric Raphaël
  • Ware Cédric
  • Jaouën Yves
  , 2020, pp.M4A.1. (10.1364/OFC.2020.M4A.1)
  DOI : 10.1364/OFC.2020.M4A.1
• Exploiting the stimuli encoding scheme of evolving Spiking Neural Networks for stream learning
  
  • Lobo Jesus L.
  • Oregi Izaskun
  • Bifet Albert
  • Ser Javier Del
  Neural Networks, Elsevier, 2020, 123, pp.118--133. Stream data processing has lately gained momentum with the arrival of new Big Data scenarios and applications dealing with continuously produced information flows. Unfortunately, traditional machine learning algorithms are not prepared to tackle the specific challenges imposed by data stream processing, such as the need for learning incrementally, limited memory and processing time requirements, and adaptation to non-stationary data, among others. To face these paradigms, Spiking Neural Networks have emerged as one of the most promising stream learning techniques, with variants such as Evolving Spiking Neural Networks capable of efficiently addressing many of these challenges. Interestingly, these networks resort to a particular population encoding scheme – Gaussian Receptive Fields – to transform the incoming stimuli into temporal spikes. The study presented in this manuscript sheds light on the predictive potential of this encoding scheme, focusing on how it can be applied as a computationally lightweight, model-agnostic preprocessing step for data stream learning. We provide informed intuition to unveil under which circumstances the aforementioned population encoding method yields effective prediction gains in data stream classification with respect to the case where no preprocessing is performed. Results obtained for a variety of stream learning models and both synthetic and real stream datasets are discussed to empirically buttress the capability of Gaussian Receptive Fields to boost the predictive performance of stream learning methods, spanning further research towards extrapolating our findings to other machine learning problems. (10.1016/J.NEUNET.2019.11.021)
  DOI : 10.1016/J.NEUNET.2019.11.021
• Experiments on Cloud-RAN Wireless Handover using Optical Switching in a Dense Urban Testbed
  
  • Minakhmetov Artur
  • Gutterman Craig
  • Chen Tingjun
  • Yu Jiakai
  • Ware Cédric
  • Iannone Luigi
  • Kilper Daniel
  • Zussman Gil
  , 2020, pp.Th2A.25. (10.1364/OFC.2020.Th2A.25)
  DOI : 10.1364/OFC.2020.Th2A.25
• La couche physique : un élément-clé des réseaux sans fil
  
  • Ciblat Philippe
  • Sibille Alain
  Annales des Mines - Enjeux Numériques, Conseil général de l'Économie, ministère de l'Économie et des Finances, 2020.
• Discriminative Streaming Network Embedding
  
  • Qi Yiyan
  • Cheng Jiefeng
  • Chen Xiaojun
  • Cheng Reynold
  • Bifet Albert
  • Wang Pinghui
  Knowledge-Based Systems, Elsevier, 2020, 190, pp.105138. Many real-world networks (e.g., friendship network among Facebook users) generate data (e.g., friend requests) in a stream fashion. Recently, several network embedding methods are proposed to learn embeddings on such networks incrementally. However, these methods perform incremental updates in a heuristic manner and thus fail to quantitatively restrict the differences between incremental learning and direct learning on the entire network (i.e., the batch learning). Moreover, they ignore the node labels (e.g., interests) when learning node embeddings, which undermines the performance of network embeddings for applications such as node classification. To solve this problem, in this paper we propose a novel network embedding framework, Discriminative Streaming Network Embedding (DimSim). When an edge insertion/deletion occurs, DimSim fast learns node embeddings incrementally, which is desired for many online applications such as anomaly detection. With the incremental learning method, at any time, the objective function well approximates to that of batch learning on the current snapshot. More importantly, the average amount of updating operations of DimSim for processing each newly coming edge is about Many real-world networks (e.g., friendship network among Facebook users) generate data (e.g., friend requests) in a stream fashion. Recently, several network embedding methods are proposed to learn embeddings on such networks incrementally. However, these methods perform incremental updates in a heuristic manner and thus fail to quantitatively restrict the differences between incremental learning and direct learning on the entire network (i.e., the batch learning). Moreover, they ignore the node labels (e.g., interests) when learning node embeddings, which undermines the performance of network embeddings for applications such as node classification. To solve this problem, in this paper we propose a novel network embedding framework, Discriminative Streaming Network Embedding (DimSim). When an edge insertion/deletion occurs, DimSim fast learns node embeddings incrementally, which is desired for many online applications such as anomaly detection. With the incremental learning method, at any time, the objective function well approximates to that of batch learning on the current snapshot. (10.1016/J.KNOSYS.2019.105138)
  DOI : 10.1016/J.KNOSYS.2019.105138
• Model-Based Virtual Prototyping of CPS: Application to Bio-Medical Devices
  
  • Genius Daniela
  • Bournias Ilias
  • Apvrille Ludovic
  • Chotin Roselyne
  , 2021, 1361, pp.74-96. Virtual prototyping and co-simulation of mixed analog/ digital embedded systems have emerged as a promising research topic, in particular for designing medical appliances. In the paper, we show how the integration of different, analog and digital, Models of Computation (MoC) within an UML/SysML based environment, can offer an efficient assistance for designing a cyber-physical system in a progressive and systematic manner. For this, we rely on formal verification and abstract simulation on a high abstraction level, and on Multi-MoC virtual prototyping on a lower abstraction level. A realistic echo monitoring system illustrates (i) the method, (ii) the modeling languages, and (iii) the different verification techniques. (10.1007/978-3-030-67445-8_4)
  DOI : 10.1007/978-3-030-67445-8_4