Publications

2019

• Fundamental limits of high bit rate optical communication
  
  • Gallion Philippe
  , 2019.
• How unitizing affects annotation of cohesion
  
  • Ceccaldi Eleonora
  • Lehmann-Willenbrock Nale
  • Volta Erica
  • Chetouani Mohamed
  • Volpe Gualtiero
  • Varni Giovanna
  , 2019, pp.1-7. (10.1109/ACII.2019.8925527)
  DOI : 10.1109/ACII.2019.8925527
• Method for Identifying Individual Electronic Devices Focusing on Differences in Spectrum Emissions
  
  • Kaji Shugo
  • Kinugawa Masahiro
  • Fujimoto Daisuke
  • Sauvage Laurent
  • Danger Jean-Luc
  • Hayashi Yuichi
  , 2019.
• Slices of Attention in Asynchronous Video Job Interviews
  
  • Hemamou Léo
  • Felhi Ghazi
  • Martin Jean-Claude
  • Clavel Chloé
  , 2019. The impact of non verbal behaviour in a hiring decision remains an open question. Investigating this question is important, as it could provide a better understanding on how to train candidates for job interviews and make recruiters be aware of influential non verbal behaviour. This research has recently been accelerated due to the development of tools for the automatic analysis of social signals (facial expression detection, speech processing, etc), and the emergence of machine learning methods. However, these studies are still mainly based on hand engineered features, which imposes a limit to the discovery of influential social signals. On the other side, deep learning methods are a promising tool to discover complex patterns without the necessity of feature engineering. In this paper, we focus on studying influential non verbal social signals in asynchronous job video interviews that are discovered by deep learning methods. We use a previously published deep learning system that aims at inferring the hirability of a candidate with regard to a sequence of interview questions. One particularity of this system is the use of attention mechanisms, which aim at identifying the relevant parts of an answer. Thus, information at a fine-grained temporal level could be extracted using global (at the interview level) annotations on hirability. While most of the deep learning systems use attention mechanisms to offer a quick visualization of slices when a rise of attention occurs, we perform an in-depth analysis to understand what happens during these moments. First, we propose a methodology to automatically extract slices where there is a rise of attention (attention slices). Second, we study the content of attention slices by comparing them with randomly sampled slices. Finally, we show that they bear significantly more information for hirability than randomly sampled slices, and that such information is related to visual cues associated with anxiety and turn taking. (10.1109/ACII.2019.8925439)
  DOI : 10.1109/ACII.2019.8925439
• Statistical Modeling of the Patches DC Component for Low-Frequency Noise Reduction
  
  • Houdard Antoine
  , 2019. In this work, we consider an additive white Gaussian noise (AWGN) model on the image patches in the context of patch-based image denoising. From this, we propose a derivation of the induced models on the centered patch of noise and on the DC component of the noise. These models allow us to treat separately the two component. We provides experiments with the HDMI method [1] that lead to denoising quality improvements, particularly for residual low frequency noise. (10.23919/EUSIPCO.2019.8903131)
  DOI : 10.23919/EUSIPCO.2019.8903131
• Cauchy Multichannel Speech Enhancement with a Deep Speech Prior
  
  • Fontaine Mathieu
  • Nugraha Aditya Arie
  • Badeau Roland
  • Yoshii Kazuyoshi
  • Liutkus Antoine
  , 2019. We propose a semi-supervised multichannel speech enhancement system based on a probabilistic model which assumes that both speech and noise follow the heavy-tailed multi-variate complex Cauchy distribution. As we advocate, this allows handling strong and adverse noisy conditions. Consequently, the model is parameterized by the source magnitude spectrograms and the source spatial scatter matrices. To deal with the non-additivity of scatter matrices, our first contribution is to perform the enhancement on a projected space. Then, our second contribution is to combine a latent variable model for speech, which is trained by following the variational autoencoder framework, with a low-rank model for the noise source. At test time, an iterative inference algorithm is applied, which produces estimated parameters to use for separation. The speech latent variables are estimated first from the noisy speech and then updated by a gradient descent method, while a majorization-equalization strategy is used to update both the noise and the spatial parameters of both sources. Our experimental results show that the Cauchy model outperforms the state-of-art methods. The standard deviation scores also reveal that the proposed method is more robust against non-stationary noise. (10.23919/EUSIPCO.2019.8903091)
  DOI : 10.23919/EUSIPCO.2019.8903091
• Impact of Intentional Electromagnetic Interference on Pure Combinational Logic
  
  • Trabelsi Oualid
  • Sauvage Laurent
  • Danger Jean-Luc
  , 2019, pp.398-403. Electromagnetic fault injection is a growing topic when it is applied to jeopardize the security of integrated circuit. Indeed, if the main part of the process will focus on the hardware efficiency of the near-field probes, tweaking properties of the electromagnetic disturbance can also lead to the success of the attack. In this paper, we are presenting characterization results of intentional electromagnetic interference by measuring its impact within the target, and more precisely on the propagation delay of a combinational logic path. The evaluation of the impact shows that the electromagnetic coupling between the probe and the integrated circuit strongly depends on the characterized properties.
• Learning Methods for RSSI-based Geolocation: A Comparative Study
  
  • Elgui Kevin
  • Bianchi Pascal
  • Portier François
  • Isson Olivier
  , 2019. In this paper, we investigate machine learning approaches addressing the problem of geolocation. First, we review some classical learning methods to build a radio map. In particular, these methods are splitted in two categories, which we refer to as likelihood-based methods and fingerprinting methods. Then, we provide a novel geolocation approach in each of these two categories. The first proposed technique relies on a semi-parametric Nadaraya-Watson estimator of the likelihood, followed by a maximum a posteriori (MAP) estimator of the object's position. The second technique consists in learning a proper metric on the dataset, constructed by means of a Gradient boosting regressor: a k-nearest neighbor algorithm is then used to estimate the position. Finally, all the proposed methods are compared on a data set originated from Sigfox network. The experiments show the interest of the proposed methods, both in terms of location estimation performance, and of ability to build radio maps.
• Regression to a Linear Lower Bound With Outliers: An Exponentially Modified Gaussian Noise Model
  
  • Gori Julien
  • Rioul Olivier
  , 2019. A regression method to estimate a linear bound in the presence of outliers is discussed. An exponentially-modified Gaussian (EMG) noise model is proposed, based on a maximum entropy argument. The resulting "EMG regression" method is shown to encompass the classical linear regression (with Gaussian noise) and a minimum regression (with exponential noise) as special cases. Simulations are performed to assess the consistency of the regression as well as its resilience to model mismatch. We conclude with an example taken from a real-world study of human performance in rapid aiming with application to human-computer interaction.
• Improving singing voice separation using Deep U-Net and Wave-U-Net with data augmentation
  
  • Cohen-Hadria Alice
  • Roebel Axel
  • Peeters Geoffroy
  , 2019, pp.1-5. (10.23919/EUSIPCO.2019.8902810)
  DOI : 10.23919/EUSIPCO.2019.8902810
• The Sliding Frank-Wolfe Algorithm for the BLASSO
  
  • Denoyelle Quentin
  • Duval Vincent
  • Peyré Gabriel
  • Soubies Emmanuel
  , 2019 (paper no. 172). This paper showcases the theoretical and numerical performance of the Sliding Frank-Wolfe, which is a novel optimization algorithm to solve the BLASSO sparse spikes super-resolution problem. The BLASSO is a continuous (i.e. off-the-grid or grid-less) counterpart to the well-known 1 sparse regularisation method (also known as LASSO or Basis Pursuit). Our algorithm is a variation on the classical Frank-Wolfe (also known as conditional gradient) which follows a recent trend of interleaving convex optimization updates (corresponding to adding new spikes) with non-convex optimization steps (corresponding to moving the spikes). Our main theoretical result is that this algorithm terminates in a finite number of steps under a mild non-degeneracy hypothesis. We then target applications of this method to several instances of single molecule fluorescence imaging modalities, among which certain approaches rely heavily on the inversion of a Laplace transform. Our second theoretical contribution is the proof of the exact support recovery property of the BLASSO to invert the 1-D Laplace transform in the case of positive spikes. On the numerical side, we conclude this paper with an extensive study of the practical performance of the Sliding Frank-Wolfe on different instantiations of single molecule fluorescence imaging, including convolutive and non-convolutive (Laplace-like) operators. This shows the versatility and superiority of this method with respect to alternative sparse recovery technics.
• Robust license plate recognition using neural networks trained on synthetic images
  
  • Björklund Tomas
  • Fiandrotti Attilio
  • Annarumma Mauro
  • Francini Gianluca
  • Magli Enrico
  Pattern Recognition, Elsevier, 2019, 93, pp.134-146. (10.1016/j.patcog.2019.04.007)
  DOI : 10.1016/j.patcog.2019.04.007
• Reliability analysis of NAND-like spintronic memory
  
  • Maciel N.
  • C. Marques E.
  • Naviner Lirida
  • Cai H.
  • Yang J.
  Microelectronics Reliability, Elsevier, 2019, 100-101, pp.113332. This work deals with magnetic tunnel junction (MTJ) based non-volatile memories (NVMs) and focuses on radiation issues. We investigate the Single Event Transient (SET) effects on a 28-nm FDSOI NAND-SPIN. The objective is to check the influence of the common current path in the metal strip of NAND-SPIN shared among various MTJs in terms of SET impact. The critical charge is determined according to the SET striking time and hardening strategies that improve the write operation robustness against SETs are proposed. (10.1016/j.microrel.2019.06.024)
  DOI : 10.1016/j.microrel.2019.06.024
• Resource Optimization for Cognitive Satellite Systems with Incumbent Terrestrial Receivers
  
  • Louchart Arthur
  • Ciblat Philippe
  • de Kerret Paul
  , 2019. We address the resource optimization issue for communications from terrestrial users to a multi-beam satellite when the bandwidth is shared with incumbent primary terrestrial systems. As a consequence, the satellite system is limited by interference temperature in order not to disturb the incumbent systems. Compared to the state of the art, we propose a relevant way to manage the interference constraints on the incumbent systems. Simulations exhibit a substantial gain in data rate when the number of incumbent systems grows.
• How to Dimension 5G Network When Users Are Distributed on Roads Modeled by Poisson Line Process
  
  • Rachad Jalal
  • Nasri Ridha
  • Decreusefond Laurent
  , 2019. The fifth generation (5G) New Radio (NR) interface inherits many concepts and techniques from 4G systems such as the Orthogonal Frequency Division Multiplex (OFDM) based waveform and multiple access. Dimensioning 5G NR interface will likely follow the same principles as in 4G networks. It aims at finding the number of radio resources required to carry a forecast data traffic at a target users Quality of Services (QoS). The present paper attempts to provide a new approach of dimension-ing 5G NR radio resource (number of Physical Resource Blocks) considering its congestion probability, qualified as a relevant metric for QoS evaluation. Moreover, 5G users are assumed to be distributed in roads modeled by Poisson Line Process (PLP) instead of the widely-used 2D-Poisson Point Process. We derive the analytical expression of the congestion probability for analyzing its behavior as a function of network parameters. Then, we set its value, often targeted by the operator, in order to find the relation between the necessary resources and the forecast data traffic expressed in terms of cell throughput. Different numerical results are presented to justify this dimensioning approach.
• Allocation de ressources pour des réseaux satellitaires cognitifs
  
  • Louchart Arthur
  • de Kerret Paul
  • Ciblat Philippe
  , 2019. Nous traitons le problème d'allocation de ressources dans une liaison montante entre des utilisateurs terrestres et un satellite qui couvre plusieurs faisceaux lorsque les bandes utilisées sont déjà allouées à un système primaire terrestre. Par conséquent, le système à optimiser ne doit pas dépasser des seuils d'interférence sur le système primaire. Par rapport à l'état de l'art, l'originalité du travail réside dans la ges-tion performante des contraintes d'interférence avec le système primaire. Nos simulations montrent que notre proposition est particulièrement avantageuse lorsque le nombre de systèmes primaires présent est élevé. Abstract-We address the resource optimization issue for communications from terrestrial users to a multi-beam satellite when the bandwidth is shared with incumbent primary terrestrial systems. As a consequence, the satellite system is limited by a interference temperature in order not to disturb the incumbent systems. Compared to the state of the art, we propose a relevant way to manage the interference constraints on the incumbent systems. Simulations exhibit a substantial gain in data rate when the number of incumbent systems grows.
• Change in Continuity: Chaining Services with an Augmented IGP
  
  • Wion Adrien
  • Bouet Mathieu
  • Iannone Luigi
  • Conan Vania
  IEEE Transactions on Network and Service Management, IEEE, 2019, pp.1-1. (10.1109/TNSM.2019.2944011)
  DOI : 10.1109/TNSM.2019.2944011
• Imagine R...
  
  • Zayana Karim
  • Rioul Olivier
  Bulletin de l’Union des Professeurs de classes préparatoires Scientifiques, 2019, 92 (267), pp.30-45. La construction de R en tant que corps totalement ordonné complet est un tour de force. Jugée complexe, elle sert pourtant de socle à une multitude de propriétés couramment utilisées. Nous en traduisons plusieurs grâce à l'obser-vation attentive des nombres et de leurs décimales. Cette volonté première à visée pédagogique nous conduit progressivement à revisiter plusieurs résultats, puis à en présenter de nouveaux qui donnent aussi matière à cet article.
• Lower Bound on Frequency Validity of Energy-Stress Tensor Based Diffuse Sound Field Model
  
  • Meacham Aidan
  • Badeau Roland
  • Polack Jean-Dominique
  , 2019. A lower bound on the frequency validity limit is established for an energetic wave equation derived from the energy-stress tensor, examined in the one-dimensional case [Dujourdy et al, Acta Acustica united with Acustica 103:480-491, 2017]. The method efficiently models diffuse sound fields that dominate reverberation at higher frequencies and larger distances. Initially noted in the course of an exhaustive search of the solution space of all valid model parameters, the low-frequency cutoff has implications for the utility of the method in a hybridization context. In practice, the bound is encountered when determining the absorption and diffusion coefficients by iteratively approaching the temporal and spatial decay of measured data. As the test frequency decreases, the ranges of coefficient combinations that result in less than 10% variation from each decay measure can diverge until the region where both measures are satisfactory (the intersection of the two domains) disappears. Further evidence for the bound is provided through comparison with measurements of a long hallway, and stability concerns in the cases where both coefficients are very small are addressed. (10.18154/RWTH-CONV-239324)
  DOI : 10.18154/RWTH-CONV-239324
• Polynomial direct sum masking to protect against both SCA and FIA
  
  • Carlet Claude
  • Daif Abderrahman
  • Guilley Sylvain
  • Tavernier Cédric
  Journal of Cryptographic Engineering, Springer, 2019, 9 (3), pp.303-312. (10.1007/s13389-018-0194-9)
  DOI : 10.1007/s13389-018-0194-9
• A meta-analysis of 1,119 manipulative experiments on terrestrial carbon-cycling responses to global change
  
  • Song Jian
  • Wan Shiqiang
  • Piao Shilong
  • Knapp Alan
  • Classen Aimée T
  • Vicca Sara
  • Ciais Philippe
  • Hovenden Mark
  • Leuzinger Sébastian
  • Beier Claus
  • Kardol Paul
  • Xia Jianyang
  • Liu Qiang
  • Ru Jingyi
  • Zhou Zhenxing
  • Luo Yiqi
  • Guo Dali
  • Adam Langley J.
  • Zscheischler Jakob
  • Dukes Jeffrey
  • Tang Jianwu
  • Chen Jiquan
  • Hofmockel Kirsten
  • Kueppers Lara
  • Rustad Lindsey
  • Liu Lingli
  • Smith Melinda
  • Templer Pamela
  • Quinn Thomas R.
  • Norby Richard
  • Phillips Richard
  • Niu Shuli
  • Fatichi Simone
  • Wang Yingping
  • Shao Pengshuai
  • Han Hongyan
  • Wang Dandan
  • Lei Lingjie
  • Wang Jiali
  • Li Xiaona
  • Zhang Qian
  • Li Xiaoming
  • Su Fanglong
  • Liu Bin
  • Yang Fan
  • Ma Gaigai
  • Li Guoyong
  • Liu Yanchun
  • Liu Yinzhan
  • Yang Zhongling
  • Zhang Kesheng
  • Miao Yuan
  • Hu Mengjun
  • Yan Chuang
  • Zhang Ang
  • Zhong Mingxing
  • Hui Yan
  • Li Ying
  • Zheng Mengmei
  Nature Ecology & Evolution, Nature, 2019, 3 (9), pp.1309-1320. (10.1038/s41559-019-0958-3)
  DOI : 10.1038/s41559-019-0958-3
• Very Low Bitrate Semantic Compression of Airplane Cockpit Screen Content
  
  • Mitrica Iulia
  • Mercier Eric
  • Ruellan Christophe
  • Fiandrotti Attilio
  • Cagnazzo Marco
  • Pesquet-Popescu Béatrice
  IEEE Transactions on Multimedia, Institute of Electrical and Electronics Engineers, 2019, 21 (9), pp.2157 - 2170. This work addresses the problem of encoding the video generated by the screen of an airplane cockpit. As other computer screens, cockpit screens consists in computer-generated graphics often atop natural background. Existing screen content coding schemes fail notably in preserving the readability of textual information at the low bitrates required in avionic applications. We propose a screen coding scheme where textual information is encoded according to the relative semantics rather than in the pixel domain. The encoder localizes textual information, the semantics of each character are extracted with a convolutional neural network and are predictively encoded. Text is then removed via inpainting, the residual background video is compressed with a standard codec and transmitted to the receiver together with the text semantics. At the decoder side, text is synthesized using the decoded semantics and superimposed over the decoded residual video recovering the original frame. Our proposed scheme offers two key advantages over a semantics-unaware scheme that encodes text in the pixel domain. First, the text readability at the decoder is not compromised by compression artifacts, whereas the relative bitrate is negligible. Second, removal of high-frequency transform coefficients associated to the inpainted text drastically reduces the bitrate of the residual video. Experiments with real cockpit video sequences show BD-rate gains up to 82% and 69 % over a reference H.265/HEVC encoder and its SCC extension. Moreover, our scheme achieves quasi-errorless character recognition already at very low bitrates, whereas even HEVC-SCC needs at least 3 or 4 times more bit-rate to achieve a comparable error rate. (10.1109/TMM.2019.2900168)
  DOI : 10.1109/TMM.2019.2900168
• Integrating tractography in pelvic surgery: a proof of concept
  
  • Muller Cécile Olivia
  • Mille Eva
  • Virzi Alessio
  • Marret Jean-Baptiste
  • Peyrot Quoc
  • Delmonte Alessandro
  • Berteloot Laureline
  • Gori Pietro
  • Blanc Thomas
  • Grévent David
  • Boddaert Nathalie
  • Bloch Isabelle
  • Sarnacki Sabine
  Journal of Pediatric Surgery Case Reports, 2019, 48, pp.101268. Objective To show the interest of tractography in pelvic surgery by a demonstrative case of neurofibroma resected by robotic assisted laparoscopy. Summary background data Although diffusion tensor magnetic resonance imaging, along with tractography algorithms, is increasingly included in image guided neurosurgery methods, it is less frequently used for the peripheral nervous system such as the pelvic sacral plexus. We report one observation demonstrating the interest of such image data for pelvic surgery. Methods A 12-years-old girl with neurofibromatosis presented with a growing left pelvic neurofibroma on repeated pelvic MRI with an increased SUV (2.83) on PET scan. A diffusion tensor pelvic MRI was performed before and after robotically assisted tumoral resection, and a tractography algorithm was applied on both images. Results The pre-operative tractogram showed the nervous nature of the tumor in close contact with the left sacral plexus. Section of a nerve trunk encased in the neurofibroma was mandatory for the resection and well documented by the 3D enhanced vision provided by the robot. Post-operatively, the patient showed a slight paresthesia and dysesthesia of the left leg calf and of the plantar vault (left S2 territory), without any motor deficit. The post-operative tractogram showed a thinning of the left S2 sacral root and the disappearance of right aberrant nervous tracts. Conclusions This clinical case validates the ability of pelvic tractography to deliver a proper imaging of the sacral nervous network and emphasizes the potential usefulness of this approach in pelvic surgery management, with perspectives of image-guided surgery. (10.1016/j.epsc.2019.101268)
  DOI : 10.1016/j.epsc.2019.101268
• Wideband Analog-to-Digital Converter (ADC) design for power amplifiers linearization
  
  • Tchambake Yapti Kelly
  , 2019. Power consumption is nowadays one of the main challenges to overcome in the development of mobile communications networks. The power amplifier (PA) is the most power hungry component in base transceiver stations. The upcoming fifth generation of mobile telephony with wider communication bands and complex modulations further increases the constraints on the PA. To overcome this problem, it is common to use predistortion techniques that enable the power amplifier to operate with greater linearity and efficiency. An important constraint in the implementation of this technique is the digitization of the output of the amplifier which, due to non-linearities, spreads over a significantly wider spectrum than the initial signal, about 5 times in practice or even more. Pipeline Analog-to-Digital Converters (ADCs) are commonly used for this operation because it allows resolutions of greater than 10 bits to be obtained over a band of several tens or even hundreds of MHz. However, its high energy consumption pushes to find a better solution. The "Multi Stage Noise Band Cancellation" (MSNBC) architecture based on Delta Sigma modulators has the advantage of realizing different dynamics per subband and is thus a prime candidate for the feedback loop ADC of predistortion techniques. The purpose of this work is to demonstrate the feasibility of the MSNBC architecture that has so far only been studied at the system level. Our investigations allowed us to propose a suitable architecture to digitize a 20 MHz RF band signal with different resolutions per subband. A continuous time Zero-IF architecture with a second-order primary modulator and a fourth-order secondary modulator with 4-bit quantizers was adopted. This architecture has been implemented in a 65 nm CMOS technology. Transistor level simulations of the 2-4 MSNBC architecture simulations with an LTE test signal resulted in 84.5 dB SNDR in the main band and 29.2 dB in the adjacent band which contains the intermodulation products.
• Statistical learning based on Markovian data maximal deviation inequalities and learning rates
  
  • Clémençon Stéphan
  • Bertail Patrice
  • Ciołek Gabriela
  Annals of Mathematics and Artificial Intelligence, Springer Verlag, 2019, 88, pp.735–757. In statistical learning theory, numerous works established non-asymptotic bounds assessing the generalization capacity of empirical risk minimizers under a large variety of complexity assumptions for the class of decision rules over which optimization is performed, by means of sharp control of uniform deviation of i.i.d. averages from their expectation, while fully ignoring the possible dependence across training data in general. It is the purpose of this paper to show that similar results can be obtained when statistical learning is based on a data sequence drawn from a (Harris positive) Markov chain X, through the running example of estimation of minimum volume sets (MV-sets) related to X’s stationary distribution, an unsupervised statistical learning approach to anomaly/novelty detection. Based on novel maximal deviation inequalities we establish, using the regenerative method, learning rate bounds that depend not only on the complexity of the class of candidate sets but also on the ergodicity rate of the chain X, expressed in terms of tail conditions for the length of the regenerative cycles. In particular, this approach fully tailored to Markovian data permits to interpret the rate bound results obtained in frequentist terms, in contrast to alternative coupling techniques based on mixing conditions: the larger the expected number of cycles over a trajectory of finite length, the more accurate the MV-set estimates. Beyond the theoretical analysis, this phenomenon is supported by illustrative numerical experiments. (10.1007/s10472-019-09670-6)
  DOI : 10.1007/s10472-019-09670-6