HAL - Lille 3
Not a member yet
5967 research outputs found
Sort by
Diagnosis and diagnosability analysis of labeled Petri nets using reduction rules
No full textInternational audienc
Binary Control Design for a Class of Bilinear Systems: Application to a Multilevel Power Converter
No full textInternational audienceThis brief studies the local stabilization of bilinear systems with binary inputs. A novel convex optimization procedure for the robust design of switching surfaces is proposed using convex optimization tools and linear matrix inequalities. The procedure takes into account the robustness to parametric uncertainties in the system matrices. An experimental illustration for the stabilization of a three-cell multilevel power converter is given
Potential gains from specialization and diversification further to the reorganization of activities
No full textInternational audienceIn economic activities, two main forces guide firm and market structures: specialization and diversification. This paper provides new insights on this topic. We propose measuring gains due to simulated division and/or merger processes of firms. Potential gains come from a reorganization of activities through specialization/diversification and/or size effects. From a database of French farms, our findings demonstrate that even if both processes are beneficial for farming systems, the division gains outweigh the gains obtained by a merger. Moreover, mix changes are more important following a division than following a merger, implying more specialization gains than diversification gains
A policymaker's guide to a Euro area stabilization fund
No full textInternational audienceIn 2012, several reports were published advocating the creation of a genuine fiscal capacity at the European level, to strengthen the foundations of the Economic and Monetary Union (EMU) by helping member countries to adjust to asymmetric shocks. The design of a common stabilization fund should be well thought, though: what would be stabilized? When would transfers be paid? Would they be directed towards governments or individuals? How would they be financed? And would they help stabilize economies? We address these questions by outlining recent existing proposals, their features and their simulated stabilization effects. We finally underline the difficulties in implementing such a scheme
A dirty deed done dirt cheap: Reporting the blame of a national reform on local politicians
No full textInternational audienceThis paper tests the hypothesis that upper-level governments can transfer the accountability of the costs of a reform to a lower one. The reform of the school week in France provides the ground for a verification of the attribution of accountability hypothesis, as it was nationally decided and locally implemented, right before a municipal election. The results confirm that local incumbents have taken the blame of the reform, especially in larger cities. In this case, thus, the cost of the reform is borne twice by the lower level of government, financially and politically. So doing, the central government does a dirty deed to the local ones, for a very cheap cost. That mayors who have announced a boycott of the reform have received electoral gains confirms that some local politicians expected to be the fall guys, bearing the brunt of the costs of the reform
In-plane face orientation estimation in still images
No full textInternational audienceThis paper addresses a fine in-plane (roll) face orientation estimation for a perspective face analysis algorithm that requires normalized frontal faces. As most of the face analysers (e.g., gender, expression, and recognition) need frontal up-right faces, there is a clear need for the precise roll estimation, as precise face normalization has an important role in classification methods. The in-plane orientation estimation algorithm is constructed on top of regular Viola-Jones frontal face detector. When a face is detected for the first time, it is rotated with respect to the face origin to find the boundaries of the detection. Mean value of these angles is said to be the measurement of the in-plane rotation of the face. Since we only need a face detection algorithm, the proposed method can work effectively on very small sized faces where traditional landmark (eye, mouth) or planar detection based estimations fail. Experiments on controlled and unconstrained large-scale datasets (CMU Rotated, YouTube, Boston University Face Tracking, Caltech, FG-NET Aging, BioID and Manchester Talking-Face) showed that the proposed method is robust to various settings for in-plane face orientation estimation in terms of RMSE and MAE. We achieved less than ±3.5 ∘ mean absolute error for roll estimation which proves that the accuracy of the proposed method is comparable to that of the state-of-the-art tracking based approaches for the roll estimation
Un algorithme de Gossip pour l’optimisation décentralisée de fonctions sur les paires
No full textInternational audienc
Les Learning Centres, un nouveau modèle d’accès et de production dessavoirs
No full textInternational audienceL'évolution en cours des bibliothèques universitaires et scolaires en Learning Centres modifie la conception de la relation entre documentation-bibliothèque et apprentissage-formation. Dans un contexte marqué par la mutation des pratiques socio-culturelles et le développement des dispositifs numériques, nous observons les formes d'appropriation et les modes d'instrumentation par les différents acteurs de ce nouveau modèle de bibliothèque. La dimension pédagogique et recherche de ces espaces documentaires est accentuée au profit de la circulation de l'information et de la mutualisation des connaissances. De quelle manière le numérique modifie-t-il les modalités de production et de diffusion des savoirs
Compter les globules blancs, analyser les partitions
No full textResearch in text algorithmics focuses on comparing, analyzing, handling or indexing sequences of characters. Working since 2006 as a CNRS researcher in computer science in the LIFL laboratory, now CRIStAL (UMR 9189, Université de Lille), Mathieu Giraud plays with sequence comparisons. This habilitation thesis focuses on two projects he worked on the last five years. In the Bonsai team, joint with Inria, Mathieu leads with Mikaël Salson a project on bioinformatics applied to hematology and immunology on the analysis of lymphocytes through their V(D)J recombinations ("Counting white blood cells"). This project, started by a collaboration with bioinformaticians and hematologists in the Lille hospital, combines research in algorithmics for immunology and hematology, software engineering and fundamental and clinical applications. The Vidjil software developed by Mathieu and his colleagues is now used by several labs in France and worldwide, sometimes in a routine hospital practice as in the Lille hospital.Mathieu also leads a projet on computer music ("Analyzing music scores"). Digital humanities link computational methods to cultural heritage and humanities research. Can computers understand music? The Algomus emergent team, shared between the CRIStAL and the MIS (Université de Picardie Jules Verne, Amiens) labs, combines musicological knowledge and computer science methods to propose algorithms analyzing music scores -- with focus on patterns, chords and chord progressions, music texture and high-level structure of music. Algomus collaborates with music theorists, music teachers and artists, and contributes to science and art projects. This thesis concludes by the description of some popular science events.Comparer, et plus généralement traiter, analyser ou indexer les séquences de caractères constitue le champ de recherche de l''algorithmique du texte. Chercheur CNRS en informatique depuis fin 2006 dans le laboratoire LIFL, maintenant CRIStAL (UMR 9189, Université de Lille), Mathieu Giraud a eu son parcours académique rythmé par les comparaisons de séquences. Ce manuscrit d'habilitation décrit les deux projets dans lesquels il s'est investi les cinq dernières années.Au sein de l'équipe Bonsai, commune avec le centre Inria Lille, Mathieu mène avec Mikaël Salson un projet de bioinformatique appliqué à l'hématologie et l'immunologie sur l'analyse des populations de lymphocytes par leurs recombinaisons V(D)J (« Compter les globules blancs »). Débuté par une collaboration avec des collègues bioinformaticiens et hématologues de l'hôpital de Lille, ce projet combine algorithmique pour l'immunologie et l'hématologie, développement logiciel et applications fondamentales et cliniques. Le logiciel Vidjil conçu par Mathieu et ses collègues est utilisé régulièrement par plusieurs laboratoires en France et à l'étranger, dont, en situation de routine, à l'hôpital de Lille.Mathieu dirige aussi un projet d'informatique musicale (« Analyser les partitions »). Les humanités numériques lient les méthodes informatiques au patrimoine culturel et à la recherche en sciences humaines et sociales. Est-ce qu'un ordinateur peut comprendre la musique? L'équipe émergente Algomus, répartie entre les laboratoires MIS (Amiens, Univ. Picardie Jules Verne) et CRIStAL, rassemble expertise musicologique et compétences algorithmiques pour proposer des méthodes analysant les partitions musicales -- comparaison de motifs et d'accords, détection de textures, analyse de formes. Algomus mène des collaborations pluridisciplinaires avec des musicologues, des professeurs de musique et des artistes et réalise des projets combinant science et art.Ce manuscrit se conclut par la description d'actions de médiation scientifique et artistique
