Publications

2018

• From Transductive to Inductive Semi-Supervised Attributes for Ship Category Recognition
  
  • Oliveau Quentin
  • Sahbi Hichem
  , 2018, pp.4827-4830. Fine-grained ship category recognition is a data-hungry learning task that requires a lot of labeled data which are usually scarce. Alternative models, as transductive attributes, bypass this limitation by considering not only labeled data but also abundant unlabeled ones. However, these transductive methods are basically designed for observed data, and their extension to unobserved sets requires retraining the whole models. In this paper, we introduce a novel ship category recognition method based on semi-supervised learning; the strength of our method resides in its ability to leverage labeled and unlabeled observed data while being highly effective and efficient in order to handle unobserved ones. We consider two variants of our method, the first one is non-parametric and based on support vector regression while the second one is parametric and based on deep neural networks. Experiments conducted on the challenging fine-grained ship category recognition show that our semi-supervised method is highly effective and generalizes well across unobserved sets. (10.1109/IGARSS.2018.8518265)
  DOI : 10.1109/IGARSS.2018.8518265
• Subsampling for big data : some recent advances
  
  • Bertail Patrice
  • Jelassi Ons
  • Tressou Jessica
  • Zetlaoui Mélanie
  , 2018.
• A Simple and Exact Algorithm to Solve l1 Linear Problems: Application to the Compressive Sensing Method
  
  • Ciril Igor
  • Darbon Jérôme
  • Tendero Yohann
  , 2018, 4, pp.54-62. This paper considers l1-regularized linear inverse problems that frequently arise in applications. One striking example is the so called compressive sensing method that proposes to reconstruct a high dimensional signal u from low dimensional measurements b=Au. The basis pursuit is another example. For most of these problems the number of unknowns is very large. The recovered signal is obtained as the solution to an optimization problem and the quality of the recovered signal directly depends on the quality of the solver. Theoretical works predict a sharp transition phase for the exact recovery of sparse signals. However, to the best of our knowledge, other state-of-the-art algorithms are not effective enough to accurately observe this transition phase. This paper proposes a simple algorithm that computes an exact l1 minimizer under the constraints Au=b. This algorithm can be employed in many problems: as soon as A has full row rank. In addition, a numerical comparison with standard algorithms available in the iterature is exhibited. These comparisons illustrate that our algorithm compares advantageously: the aforementioned transition phase is empirically observed with a much better quality. (10.5220/0006624600540062)
  DOI : 10.5220/0006624600540062
• Efficient Bayesian Computation by Proximal Markov Chain Monte Carlo: When Langevin Meets Moreau.
  
  • Durmus Alain
  • Moulines Éric
  • Pereyra Marcelo
  SIAM Journal on Imaging Sciences, Society for Industrial and Applied Mathematics, 2018, 11 (1). In this paper, two new algorithms to sample from possibly non-smooth log-concave probability measures are introduced. These algorithms use Moreau-Yosida envelope combined with the Euler-Maruyama discretization of Langevin diffusions. They are applied to a de-convolution problem in image processing, which shows that they can be practically used in a high dimensional setting. Finally, non-asymptotic bounds for one of the proposed methods are derived. These bounds follow from non-asymptotic results for ULA applied to probability measures with a convex continuously differentiable log-density with respect to the Lebesgue measure. (10.1137/16M110834)
  DOI : 10.1137/16M110834
• Unsupervised real-time detection of BGP anomalies leveraging high-rate and fine-grained telemetry data
  
  • Putina Andrian
  • Barth Steven
  • Bifet Albert
  • Pletcher Drew
  • Precup Cristina
  • Nivaggioli Patrice
  • Rossi Dario
  , 2018, pp.1-2. (10.1109/INFCOMW.2018.8406838)
  DOI : 10.1109/INFCOMW.2018.8406838
• Chromatic dispersion, nonlinear parameter and modulation formats monitoring based on Godard’s error for coherent optical transmission systems
  
  • Jiang Lin
  • Yan Lianshan
  • Yi Anlin
  • Pan Yan
  • Hao Ming
  • Pan Wei
  • Luo Bin
  • Jaouën Yves
  IEEE Photonics Journal, Institute of Electrical and Electronics Engineers (IEEE), 2018, 10 (1), pp.790051. This paper considers Godard's error as signal quality metric to monitor chromatic dispersion (CD), nonlinear parameter and modulation format in the DSP module of the coherent receivers. We first review a CD monitoring based on Godard's error that can be able to accurately monitor arbitrarily large dispersion values in uncompensated transmission links in combination with frequency domain equalizer, then extend the previous nonlinear parameter monitoring method based on Godard's error by blindly obtaining the optimized value γξp to significantly improve the adaptive capability, and present a simple and effective modulation format monitoring based on Godard's error. Meanwhile, the effectiveness has been experimentally verified in 128-Gb/s PDM-QPSK, 192-Gb/s PDM-8QAM, and 256-Gb/s PDM-16QAM systems. (10.1109/JPHOT.2017.2786697)
  DOI : 10.1109/JPHOT.2017.2786697
• Defining services and service orchestrators acting on shared sensors and actuators
  
  • Baghli Rayhana
  • Najm Elie
  • Traverson Bruno
  , 2018.
• Depiction of the perfusion components’ volume fraction distribution in generalized intravoxel incoherent motion by using Gaussian mixture model
  
  • Wang Shunli
  • Liu Wanyu
  • Kuai Zixiang
  • Zhu Yuemin
  Concepts in Magnetic Resonance Part B: Magnetic Resonance Engineering, Wiley, 2018, 48 (3), pp.e21399. Gaussian mixture model (GMM) was proposed to depict the perfusion volume fraction distribution in the generalized intravoxel incoherent motion model (GIVIM) to improve GIVIM's ability of describing complex perfusion conditions and their changes. Different hepatic perfusion conditions were accounted for by performing different combinations of imaging sequence and diffusion time on six normal livers. In order to evaluate GIVIM-GMM's reliability in perfusion condition analysis, the fitting to diffusion-weighted (DW) data and the consistency between diffusion-related parameters' change and the data's change were tested and the recent GIVIM and the triexponential models were chosen for comparison. The difference of the fitting results was evaluated by performing the extra-sum-of-squares F test and information criteria on normal human DW data. The difference of the consistency was assessed by using two-tailed paired Student's t test. In the extra-sum-of-squares F test, the relative difference ratio F values derived from theGIVIM and GIVIM-GMM and that derived from the triexponential model and the GIVIM-GMM are respectively 25.334 and 27.976, which indicates that significant difference existed and that the GIVIM-GMM provides better fit to the normal human liver DW data. In information criteria test, the evidence ratio values were determined by dividing the GIVIM's or triexponential model's correct probability by the GIVIM-GMM's. Both evidence ratio values (2.3942x10(-10), 8.6167x10(-9), respectively) are much smaller than 1, which also expresses that the best model used to fit the normal human liver DW data was the GIVIM-GMM. In two-tailed paired student's t test, the GIVIM-GMM provides more parameters to give a finer description of perfusion than the triexponential model or GIVIM. In short, all the results demonstrated that the GIVIM-GMM provides better performance than the existing IVIM models for depicting the signal attenuation in DW imaging. (10.1002/cmr.b.21399)
  DOI : 10.1002/cmr.b.21399
• Evolving Attacker Perspectives for Secure Embedded System Design
  
  • Li Letitia W.
  • Lugou Florian
  • Apvrille Ludovic
  , 2018. In our increasingly connected world, security is a growing concern for embedded systems. A systematic design and verification methodology could help detect vulnerabilities before mass production. While Attack Trees help a designer consider the attacks a system will face during a preliminary analysis phase, they can be further integrated into the design phases. We demonstrate that explicitly modeling attacker actions within a system model helps us to evaluate its impact and possible countermeasures. This paper describes how we evolved the SysML-Sec Methodology with ``Attacker Scenarios'' for the improved design of secure embedded systems.
• Mass Volume Curves and Anomaly Ranking
  
  • Clémençon Stéphan
  • Thomas Albert
  Electronic Journal of Statistics, Shaker Heights, OH : Institute of Mathematical Statistics, 2018, 12 (2). This paper aims at formulating the issue of ranking multivariate unlabeled observations depending on their degree of abnormality as an unsupervised statistical learning task. In the 1-d situation, this problem is usually tackled by means of tail estimation techniques: univariate observations are viewed as all the more `abnormal' as they are located far in the tail(s) of the underlying probability distribution. It would be desirable as well to dispose of a scalar valued `scoring' function allowing for comparing the degree of abnormality of multivariate observations. Here we formulate the issue of scoring anomalies as a M-estimation problem by means of a novel functional performance criterion, referred to as the Mass Volume curve (MV curve in short), whose optimal elements are strictly increasing transforms of the density almost everywhere on the support of the density. We first study the statistical estimation of the MV curve of a given scoring function and we provide a strategy to build confidence regions using a smoothed bootstrap approach. Optimization of this functional criterion over the set of piecewise constant scoring functions is next tackled. This boils down to estimating a sequence of empirical minimum volume sets whose levels are chosen adaptively from the data, so as to adjust to the variations of the optimal MV curve, while controling the bias of its approximation by a stepwise curve. Generalization bounds are then established for the difference in sup norm between the MV curve of the empirical scoring function thus obtained and the optimal MV curve. (10.1214/18-EJS1474)
  DOI : 10.1214/18-EJS1474
• An iterative multi-atlas patch-based approach for cortex segmentation from neonatal MRI
  
  • Tor-Díez Carlos
  • Passat Nicolas
  • Bloch Isabelle
  • Faisan Sylvain
  • Bednarek Nathalie
  • Rousseau François
  Computerized Medical Imaging and Graphics, Elsevier [1988-....], 2018, 70, pp.73-82. Brain structure analysis in the newborn is a major health issue. This is especially the case for premature neonates, in order to obtain predictive information related to the child development. In particular, the cortex is a structure of interest, that can be observed in MRI (magnetic resonance imaging). However, neonatal MRI data present specific properties that make them challenging to process. In this context, multi-atlas approaches constitute an efficient strategy, taking advantage of images processed beforehand. The method proposed in this article relies on such multi-atlas strategy. More precisely, it uses two paradigms: first, a non-local model based on patches; second, an iterative optimization scheme. Coupling both concepts allows us to consider patches related not only to the image information, but also to the current segmentation. This strategy is compared to other multi-atlas methods proposed in the literature. Experiments show that the proposed approach provides robust cortex segmentation results. (10.1016/j.compmedimag.2018.09.003)
  DOI : 10.1016/j.compmedimag.2018.09.003