HAL - Lille 3
Not a member yet
5967 research outputs found
Sort by
Formes communicationnelles et opérations sociales : une approche par les échanges au travail (des échanges en travail)
No full textInternational audienc
Introducing FoxFaces: a 3-in-1 head dataset
No full textInternational audienceWe introduce a new test collection named FoxFaces, dedicated to researchers in face recognition and analysis. The creation of this dataset was motivated by a lack encountered in the existing 3D/4D datasets. FoxFaces contains 3 face datasets obtained with several devices. Faces are captured with different changes in pose, expression and illumination. The presented collection is unique in two aspects: the acquisition is performed using three little constrained devices offering 2D, depth and stereo information on faces. In addition, it contains both still images and videos allowing static and dynamic face analysis. Hence, our dataset can be an interesting resource for the evaluation of 2D, 3D and bimodal algorithms on face recognition under adverse conditions as well as facial expression recognition and pose estimation algorithms in static and dynamic domains (images and videos). Stereo, color, and range images and videos of 64 adult human subjects are acquired. Acquisitions are accompanied with information about the subjects identity, gender, facial expression, approximate pose orientation and the coordinates of some manually located facial fiducial points
Multi-loci diagnosis of acute lymphoblastic leukaemia with high-throughput sequencing and bioinformatics analysis
No full textInternational audienceHigh-throughput sequencing (HTS) is considered a technical revolution that has improved our knowledge of lymphoid and autoimmune diseases, changing our approach to leukaemia both at diagnosis and during follow-up. As part of an immunoglobulin/T cell receptor-based minimal residual disease (MRD) assessment of acute lymphoblastic leukaemia patients, we assessed the performance and feasibility of the replacement of the first steps of the approach based on DNA isolation and Sanger sequencing, using a HTS protocol combined with bioinformatics analysis and visualization using the Vidjil software. We prospectively analysed the diagnostic and relapse samples of 34 paediatric patients, thus identifying 125 leukaemic clones with recombinations on multiple loci (TRG, TRD, IGH and IGK), including Dd2/Dd3 and Intron/KDE rearrangements. Sequencing failures were halved (14% vs. 34%, P = 0.0007), enabling more patients to be monitored. Furthermore, more markers per patient could be monitored, reducing the probability of false negative MRD results. The whole analysis, from sample receipt to clinical validation, was shorter than our current diagnostic protocol, with equal resources. V(D)J recombination was successfully assigned by the software, even for unusual recombinations. This study emphasizes the progress that HTS with adapted bioinformatics tools can bring to the diagnosis of leukaemia patients
ScapeGoat: Spotting abnormal resource usage in component-based reconfigurable software systems
No full textInternational audienceModern component frameworks support continuous deployment and simultaneous execution of multiple software components on top of the same virtual machine. However , isolation between the various components is limited. A faulty version of any one of the software components can compromise the whole system by consuming all available resources. In this paper, we address the problem of efficiently identifying faulty software components running simultaneously in a single virtual machine. Current solutions that perform permanent and extensive monitoring to detect anomalies induce high overhead on the system, and can, by themselves, make the system unstable. In this paper we present an optimistic adaptive monitoring system to determine the faulty components of an application. Suspected components are finely analyzed by the monitoring system, but only when required. Unsuspected components are left untouched and execute normally. Thus, we perform localized just-in-time monitoring that decreases the accumulated overhead of the monitoring system. We evaluate our approach on two case studies against a state-of-the-art monitoring system and show that our technique correctly detects faulty components, while reducing overhead by an average of 93%
La force du collectif
No full textNational audienceEn créant des entités informatiques capables de réagir à des événements imprévus, de dresser des plans et de travailler en équipe, les informaticiens simulent les comportements collectifs de groupes d’individus, d’animaux ou de molécules
Analyse Multi-Dimensionnelle des Consommations Energétiques Logicielles sur les Architectures Multi-Coeurs
No full textEnergy-efficient computing is becoming increasingly important. Among the reasons, one can mention the massive consumption of large data centers that consume as much as 180,000 homes. This trend, combined with environmental concerns, makes energy efficiency a prime technological and societal challenge. Currently, widely used power distribution units (PDUs) are often shared amongst nodes to deliver aggregated power consumption reports, in the range of hours and minutes. However, in order to improve the energy efficiency of software systems, we need to support process-level power estimation in real-time, which goes beyond the capacity of a PDUs.In particular, the CPU is considered by the research community as the major power consumer within a node and draws attention while trying to model the system power consumption. Over the last few years, a dozen of ad hoc power models have been proposed to cope with the wide diversity and the growing complexity of modern CPU architectures.In this thesis, we rather propose PowerAPI for learning power models and building software-defined power meters that provide accurate power estimation on modern architectures. With the emergence of cloud computing, we propose BitWatts and WattsKit for leveraging software power estimation in VMs and clusters. A finer level of estimation may be required to further evaluate the effectiveness of the software optimizations and we therefore propose codEnergy for helping developers to understand how the energy is really consumed by a software. We deeply assessed all above approaches, thus demonstrating the usefulness of PowerAPI to better understand the software power consumption on modern architectures.L’Informatique “verte” est très importante de nos jours. Parmi les raisons principales, nous pouvons mentionner le rôle majeur des centres de données qui consomment autant que 180 000 foyers en électricité. Associé aux préoccupations énergétiques, cet enjeu représente un challenge technologique et sociétal de premier ordre. Des watt-mètres sont actuellement utilisés et partagés pour récupérer un ensemble agrégé de rapports énergétiques sur plusieurs heures ou minutes. Cependant, pour améliorer l’efficacité énergétique des logiciels, nous devons dépasser ces limitations et proposer des estimations plus fines. Particulièrement, la communauté considère le CPU comme étant le composant le plus énergivore et est donc largement considéré lors de la modélisation énergétique de système. Des dizaines de modèles de consommation ont déjà été proposées pour prendre en compte la grande diversité et la complexité grandissante des CPUs. Dans cette thèse, nous proposons PowerAPI pour apprendre automatiquement les modèles de consommation et construire des watt-mètres logiciels permettant des estimations précises sur des architectures modernes. Avec l’émergence de l’Informatique dématérialisée, nous avons développé BitWatts et WattsKit pour pousser l’utilisation d’estimations énergétiques à grain fin au sein de VMs ou clusters. Un niveau encore plus fin peut être requis pour mieux évaluer l’efficacité d’optimisations logicielles et nous proposons donc codEnergy pour aider à mieux comprendre comment l’énergie est consommée par un logiciel. Nous démontrons aussi dans cette thèse l’utilité de PowerAPI pour mieux comprendre les consommations logicielles sur les architectures modernes
Stochastic model for differential Mueller matrix of stationary and non-stationary turbid media
No full textInternational audienceWe show the existence of different regimes in spatial evolution of depolarization in turbid media characterized by a diagonal Mueller matrix (pure depolarizer). Experimental results previously published already established the existence of a first regime, where the depolarization follows a parabolic law with the thickness of stationary medium traveled by light. New experiments first confirm the existence of a second regime, which we have previously demonstrated, where the depolarization follows a linear law on a large scale. They also confirm the existence of much more complex evolution laws even under small-scale approximation. A stochastic approach is proposed to model the phenomenon. It perfectly describes all these different experimental results and allows us to analyze the behavior of the polarization in the case of solid or liquid scattering media. The influence of themeasurement setup is also analyzed
Steering and strengthening knowledge economy through Open Access initiatives: Case of Zimbabwe
No full textInternational audienceThe twenty first century has seen an accelerated development in the activities of library consortia in Africa. The e-revolution has precipitated the transformation of higher education (HE) through ushering in new paradigms with regards to how information / knowledge is generated, managed and shared. It will examine Open Access in Zimbabwe with regards to generation, promotion and coordination to realise the ideals of a knowledge economy. This paper will explore the extent to which open access has been adopted in Zimbabwe's HE institutions. It will also examine the extent of use of information and communication technologies (ICTs) in widening access to scholarly communication. The writers will examine the roles of library consortia, government and research networks in the open access dispensation. It will explore the challenges and opportunities emanating from open access. The paper will examine the extent to which academic institutions have adopted institutional repositories to promote access to e-content. The paper will examine challenges relating to intellectual property rights, for example, licensing and access to open educational resources. The paper will also explore how library consortia can derive benefits from open access. The paper will examine how Open Access Initiatives can help contribute to the knowledge economy. Zimbabwe. He has keen interest in social media, indigenous knowledge systems, marketing, open access, digital libraries, changing role of librarians, and electronic information services. He presented in various conferences in Zimbabwe and in Africa
Supervision of time-frequency features selection in EEG signals by a human expert for brain–computer interfacing based on motor imagery
No full textInternational audienceIn the context of brain–computer interfacing based on motor imagery, we propose a method which allows an expert to select manually time-frequency features. This selection is performed specifically for each subject, by analysing a set of curves that emphasize differences of brain activity recorded from electroencephalographic signals during the execution of various motor imagery tasks. We will show that expert knowledge is very valuable to supervise the selection of a sparse set of significant time-frequency features. Features selection is performed through a graphical user interface to allow an easy access to experts with no specific programming skills. In this paper, we compare our method with three fully-automatic features selection methods, using dataset 2A of BCI competition IV. Results are better for five of the nine subjects compared to the best competing method