École Nationale Supérieure de Techniques Avancées Paris

<para xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> Exploratory activities seem to be intrinsically rewarding for children and crucial for their cognitive development. Can a machine be endowed with such an intrinsic motivation system? This is the question we study in this paper, presenting a number of computational systems that try to capture this drive towards novel or curious situations. After discussing related research coming from developmental psychology, neuroscience, developmental robotics, and active learning, this paper presents the mechanism of Intelligent Adaptive Curiosity, an intrinsic motivation system which pushes a robot towards situations in which it maximizes its learning progress. This drive makes the robot focus on situations which are neither too predictable nor too unpredictable, thus permitting autonomous mental development. The complexity of the robot's activities autonomously increases and complex developmental sequences self-organize without being constructed in a supervised manner. Two experiments are presented illustrating the stage-like organization emerging with this mechanism. In one of them, a physical robot is placed on a baby play mat with objects that it can learn to manipulate. Experimental results show that the robot first spends time in situations which are easy to learn, then shifts its attention progressively to situations of increasing difficulty, avoiding situations in which nothing can be learned. Finally, these various results are discussed in relation to more complex forms of behavioral organization and data coming from developmental psychology. </para>

Long QT syndrome (LQTS) is a heterogeneous inherited disorder causing syncope and sudden death from ventricular arrhythmias. A first locus for this disorder was mapped to chromosome 11p15.5. However, locus heterogeneity has been demonstrated in several families, and two other loci have recently been located on chromosomes 7q35-36 and 3p21-24. We used linkage analysis to map the locus in a 65-member family in which LQTS was associated with more marked sinus bradycardia than usual, leading to sinus node dysfunction. Linkage to chromosome 11p15.5, 7q35-36, or 3p21-24 was excluded. Positive linkage was obtained for markers located on chromosome 4q25-27. A maximal LOD score of 7.05 was found for marker D4S402. The identification of a fourth locus for LQTS confirms its genetic heterogeneity. Locus 4q25-27 is associated with a peculiar phenotype within the LQTS entity.

Localization for low cost humanoid or animal-like personal robots has to rely on cheap sensors and has to be robust to user manipulations of the robot. We present a visual localization and map-learning system that relies on vision only and that is able to incrementally learn to recognize the different rooms of an apartment from any robot position. This system is inspired by visual categorization algorithms called bag of words methods that we modified to make fully incremental and to allow a user-interactive training. Our system is able to reliably recognize the room in which the robot is after a short training time and is stable for long term use. Empirical validation on a real robot and on an image database acquired in real environments are presented.

The present work was carried out in the scope of a numerical-experimental collaborative research program, whose main objective is to understand the mechanisms of instabilities in partial cavitating flow. Experiments were conducted in the configuration of a rectangular foil located in a cavitation tunnel. Partial cavitation was investigated by multipoint wall-pressure measurements together with lift and drag measurements and numerical videos. The computations were conducted on two-dimensional hydrofoil section and are based on a single fluid model of cavitation: the liquid/vapor mixture is considered as a homogeneous fluid whose composition is regulated by a barotropic state law. The algorithm of resolution is derived from the SIMPLE approach, modified to take into account the high compressibility of the medium. Several physical features were pointed out by this joint approach. Particularly two distinct cavity self-oscillation dynamics characterized by two different frequencies (dynamics 1 and dynamics 2) were obtained experimentally and numerically at the angles of incidence of 6° and 8°. In both cases, the reentrant jet was found to be mainly responsible for the cavity breakdown. Dynamics 2 corresponds to the “classical” cavity breakdown and resulting cloud cavitation. A more complex flow pattern was evidenced for dynamics 1. In this case the growth/breakdown cycle of the cavity was observed at a lower Strouhal number (∼0.07∕0.09) than dynamics 2 (∼0.3). Moreover, the mechanism is composed of two successive steps: (i) an interaction between the reentrant jet and the cavity interface in the closure region leading to the periodic shedding of secondary cavitation clouds before the main cloud detachment occurs, and (ii) a shock wave induced by the collapse of the main cloud, which influences the growth of the residual cavity.

We present the first single-shot images of ferromagnetic, nanoscale spin order taken with femtosecond x-ray pulses. X-ray-induced electron and spin dynamics can be outrun with pulses shorter than 80 fs in the investigated fluence regime, and no permanent aftereffects in the samples are observed below a fluence of 25 mJ/cm(2). Employing resonant spatially muliplexed x-ray holography results in a low imaging threshold of 5 mJ/cm(2). Our results open new ways to combine ultrafast laser spectroscopy with sequential snapshot imaging on a single sample, generating a movie of excited state dynamics.

Accurate and up-to-date road maps are of great importance in a wide range of applications. Unfortunately, automatic road extraction from high-resolution remote sensing images remains challenging due to the occlusion of trees and buildings, discriminability of roads, and complex backgrounds. To address these problems, especially road connectivity and completeness, in this article, we introduce a novel deep learning-based multistage framework to accurately extract the road surface and road centerline simultaneously. Our framework consists of three steps: boosting segmentation, multiple starting points tracing, and fusion. The initial road surface segmentation is achieved with a fully convolutional network (FCN), after which another lighter FCN is applied several times to boost the accuracy and connectivity of the initial segmentation. In the multiple starting points tracing step, the starting points are automatically generated by extracting the road intersections of the segmentation results, which then are utilized to track consecutive and complete road networks through an iterative search strategy embedded in a convolutional neural network (CNN). The fusion step aggregates the semantic and topological information of road networks by combining the segmentation and tracing results to produce the final and refined road segmentation and centerline maps. We evaluated our method utilizing three data sets covering various road situations in more than 40 cities around the world. The results demonstrate the superior performance of our proposed framework. Specifically, our method's performance exceeded the other methods by 7% and 40% for the connectivity indicator for road surface segmentation and for the completeness indicator for centerline extraction, respectively.

A wide variety of problems in systems and control theory can be cast or recast as convex problems that involve linear matrix inequalities (LMIs). For a few very special cases there are "analytical solutions" to these problems, but in general they can be solved numerically very efficiently. In many cases the inequalities have the form of simultaneous Lyapunov or algebraic Riccati inequalities; such problems can be solved in a time that is comparable to the time required to solve the same number of Lyapunov or Algebraic Riccati equations. Therefore the computational cost of extending current control theory that is based on the solution of algebraic Riccati equations to a theory based on the solution of (multiple, simultaneous) Lyapunov or Riccati inequalities is modest. Examples include: multicriterion LQG, synthesis of linear state feedback for multiple or nonlinear plants ("multi-model control"), optimal transfer matrix realization, norm scaling, synthesis of multipliers for Popov-like analysis of systems with unknown gains, and many others. Full details can be found in the references cited.

In this paper, we discuss the scientific outcomes of the 2015 data fusion contest organized by the Image Analysis and Data Fusion Technical Committee (IADF TC) of the IEEE Geoscience and Remote Sensing Society (IEEE GRSS). As for previous years, the IADF TC organized a data fusion contest aiming at fostering new ideas and solutions for multisource studies. The 2015 edition of the contest proposed a multiresolution and multisensorial challenge involving extremely high-resolution RGB images and a three-dimensional (3-D) LiDAR point cloud. The competition was framed in two parallel tracks, considering 2-D and 3-D products, respectively. In this paper, we discuss the scientific results obtained by the winners of the 2-D contest, which studied either the complementarity of RGB and LiDAR with deep neural networks (winning team) or provided a comprehensive benchmarking evaluation of new classification strategies for extremely high-resolution multimodal data (runner-up team). The data and the previously undisclosed ground truth will remain available for the community and can be obtained at http://www.grss-ieee.org/community/technical-committees/data-fusion/2015-ieee-grss-data-fusion-contest/. The 3-D part of the contest is discussed in the Part-B paper [1].

BACKGROUND: Polymorphic tandem repeat typing is a new generic technology which has been proved to be very efficient for bacterial pathogens such as B. anthracis, M. tuberculosis, P. aeruginosa, L. pneumophila, Y. pestis. The previously developed tandem repeats database takes advantage of the release of genome sequence data for a growing number of bacteria to facilitate the identification of tandem repeats. The development of an assay then requires the evaluation of tandem repeat polymorphism on well-selected sets of isolates. In the case of major human pathogens, such as S. aureus, more than one strain is being sequenced, so that tandem repeats most likely to be polymorphic can now be selected in silico based on genome sequence comparison. RESULTS: In addition to the previously described general Tandem Repeats Database, we have developed a tool to automatically identify tandem repeats of a different length in the genome sequence of two (or more) closely related bacterial strains. Genome comparisons are pre-computed. The results of the comparisons are parsed in a database, which can be conveniently queried over the internet according to criteria of practical value, including repeat unit length, predicted size difference, etc. Comparisons are available for 16 bacterial species, and the orthopox viruses, including the variola virus and three of its close neighbors. CONCLUSIONS: We are presenting an internet-based resource to help develop and perform tandem repeats based bacterial strain typing. The tools accessible at http://minisatellites.u-psud.fr now comprise four parts. The Tandem Repeats Database enables the identification of tandem repeats across entire genomes. The Strain Comparison Page identifies tandem repeats differing between different genome sequences from the same species. The "Blast in the Tandem Repeats Database" facilitates the search for a known tandem repeat and the prediction of amplification product sizes. The "Bacterial Genotyping Page" is a service for strain identification at the subspecies level.

This context-aware middleware system facilitates diverse multimedia services in heterogeneous network environments by combining an adaptive service provisioning middleware framework with a context-aware multimedia middleware framework.

In robotic applications of visual simultaneous localization and mapping, loop-closure detection and global localization are two issues that require the capacity to recognize a previously visited place from current camera measurements. We present an online method that makes it possible to detect when an image comes from an already perceived scene using local shape information. Our approach extends the bag of visual words method used in image recognition to incremental conditions and relies on Bayesian filtering to estimate loop-closure probability. We demonstrate the efficiency of our solution by real-time loop-closure detection under strong perceptual aliasing conditions in an indoor image sequence taken with a handheld camera.

This research focuses on studying the possible role of a socially interactive robot as a tool for monitoring and encouraging cognitive activities of the elderly and/or individuals suffering from dementia. One of the aims of this work is to show the benefits of the robot's physical embodiment in human-robot social interactions. The social therapist robot tries to provide customized cognitive stimulation by playing a music game with the user. The results of the 8-month pilot study depict a more efficient, natural, and preferred interaction with the robot rather than with the simulated robot.

Generalized additive models (GAMs) are flexible non-linear regression models, which can be fitted efficiently using the approximate Bayesian methods provided by the mgcv R package. While the GAM methods provided by mgcv are based on the assumption that the response distribution is modeled parametrically, here we discuss more flexible methods that do not entail any parametric assumption. In particular, this article introduces the qgam package, which is an extension of mgcv providing fast calibrated Bayesian methods for fitting quantile GAMs (QGAMs) in R. QGAMs are based on a smooth version of the pinball loss of Koenker (2005), rather than on a likelihood function, hence jointly achieving satisfactory accuracy of the quantile point estimates and coverage of the corresponding credible intervals requires adopting the specialized Bayesian fitting framework of Fasiolo, Wood, Zaffran, Nedellec, and Goude (2021b). Here we detail how this framework is implemented in qgam and we provide examples illustrating how the package should be used in practice.

A new hydrographic climatology has been created for the continental shelf region, extending from the Labrador shelf to the Mid-Atlantic Bight. The 0.2-degree climatology combines all available observations of surface and bottom temperature and salinity collected between 1950 and 2010 along with the location, depth and date of these measurements. While climatological studies of surface and bottom temperature and salinity have been presented previously for various regions along the Canadian and U.S. shelves, studies also suggest that all these regions are part of one coherent system. This study focuses on the coherent structure of the mean seasonal cycle of surface and bottom temperature and salinity and its variation along the shelf and upper slope. The seasonal cycle of surface temperature is mainly driven by the surface heat flux and exhibits strong dependency on latitude (r≈−0.9). The amplitude of the seasonal cycle of bottom temperature is rather dependent on the depth, while the spatial distribution of bottom temperature is correlated with latitude. The seasonal cycle of surface salinity is influenced by several components, such as sea-ice on the northern shelves and river discharge in the Gulf of St. Lawrence. The bottom salinity exhibits no clear seasonal cycle, but its spatial distribution is highly correlated with bathymetry, thus Slope Water and its intrusion on the shelf can be identified by its relatively high salinity compared to shallow, fresher shelf water. Two different regimes can be identified, especially on the shelf, separated by the Laurentian Channel: advection influences the phasing of the seasonal cycle of surface salinity and bottom temperature to the north, while in the southern region, river runoff and air-sea heat flux forcing are dominant, especially over the shallower bathymetry.

Abstract This paper investigates model-order reduction methods for geometrically nonlinear structures. The parametrisation method of invariant manifolds is used and adapted to the case of mechanical systems in oscillatory form expressed in the physical basis, so that the technique is directly applicable to mechanical problems discretised by the finite element method. Two nonlinear mappings, respectively related to displacement and velocity, are introduced, and the link between the two is made explicit at arbitrary order of expansion, under the assumption that the damping matrix is diagonalised by the conservative linear eigenvectors. The same development is performed on the reduced-order dynamics which is computed at generic order following different styles of parametrisation. More specifically, three different styles are introduced and commented: the graph style, the complex normal form style and the real normal form style. These developments allow making better connections with earlier works using these parametrisation methods. The technique is then applied to three different examples. A clamped-clamped arch with increasing curvature is first used to show an example of a system with a softening behaviour turning to hardening at larger amplitudes, which can be replicated with a single mode reduction. Secondly, the case of a cantilever beam is investigated. It is shown that invariant manifold of the first mode shows a folding point at large amplitudes. This exemplifies the failure of the graph style due to the folding point on a real structure, whereas the normal form style is able to pass over the folding. Finally, a MEMS (Micro Electro Mechanical System) micromirror undergoing large rotations is used to show the importance of using high-order expansions on an industrial example.

IPIN 2019 Competition, sixth in a series of IPIN competitions, was held at the CNR Research Area of Pisa (IT), integrated into the program of the IPIN 2019 Conference. It included two on-site real-time Tracks and three off-site Tracks. The four Tracks presented in this paper were set in the same environment, made of two buildings close together for a total usable area of 1000 m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> outdoors and and 6000 m <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> indoors over three floors, with a total path length exceeding 500 m. IPIN competitions, based on the EvAAL framework, have aimed at comparing the accuracy performance of personal positioning systems in fair and realistic conditions: past editions of the competition were carried in big conference settings, university campuses and a shopping mall. Positioning accuracy is computed while the person carrying the system under test walks at normal walking speed, uses lifts and goes up and down stairs or briefly stops at given points. Results presented here are a showcase of state-of-the-art systems tested side by side in real-world settings as part of the on-site real-time competition Tracks. Results for off-site Tracks allow a detailed and reproducible comparison of the most recent positioning and tracking algorithms in the same environment as the on-site Tracks.

International audience

We show how to refine the conformal block expansion convergence estimates from arXiv:1208.6449. In doing so we find a novel explicit formula for the 3d conformal blocks on the real axis.

We seek to define sequence-based predictive criteria to identify polymorphic and hypermutable minisatellites in the human genome. Polymorphism of a representative pool of minisatellites, selected from human chromosomes 21 and 22, was experimentally measured by PCR typing in a population of unrelated individuals. Two predictive approaches were tested. One uses simple repeat characteristics (e.g., unit length, copy number, nucleotide bias) and a more complex measure, termed HistoryR, based on the presence of variant motifs in the tandem array. We find that HistoryR and percentage of GC are strongly correlated with polymorphism and, as predictive criteria, reduce by half the number of repeats to type while enriching the proportion with heterozygosity >/=0.5, from a background level of 43% to 59%. The second approach uses length differences between minisatellites in the two releases of the human genome sequence (from the public consortium and Celera). As a predictor, this similarly enriches the number of polymorphic minisatellites, but fails to identify an unexpectedly large number of these. Finally, typing of the highly polymorphic minisatellites in large families identified one new hypermutable minisatellite, located in a predicted coding sequence. This may represent the first coding human hypermutable minisatellite.

Abstract This paper is devoted to a detailed analysis of the appearance of frequency combs in the dynamics of a micro-electro-mechanical systems (MEMS) resonator featuring 1:2 internal resonance. To that purpose, both experiments and numerical predictions are reported and analysed to predict and follow the appearance of the phononic frequency comb arising as a quasi-periodic regime between two Neimark-Sacker bifurcations. Numerical predictions are based on a reduced-order model built thanks to an implicit condensation method, where both mechanical nonlinearities and electrostatic forces are taken into account. The reduced order model is able to predict a priori, i.e. without the need of experimental calibration of parameters, and in real time, i.e. by solving one or two degrees-of-freedom system of equations, the nonlinear behaviour of the MEMS resonator. Numerical predictions show a good agreement with experiments under different operating conditions, thus proving the great potentiality of the proposed simulation tool. In particular, the bifurcation points and frequency content of the frequency comb are carefully predicted by the model, and the main features of the periodic and quasi-periodic regimes are given with accuracy, underlining that the complex dynamics of such MEMS device is effectively driven by the characteristics of the 1:2 internal resonance.