HAL - Lille 3
Not a member yet
5967 research outputs found
Sort by
Semiring Rank Matrix Factorization
No full textInternational audienceRank data, in which each row is a complete or partial ranking of available items (columns), is ubiquitous. Among others, it can be used to represent preferences of users, levels of gene expression, and outcomes of sports events. It can have many types of patterns, among which consistent rankings of a subset of the items in multiple rows, and multiple rows that rank the same subset of the items highly. In this article, we show that the problems of finding such patterns can be formulated within a single generic framework that is based on the concept of semiring matrix factorisation. In this framework, we employ the max-product semiring rather than the plus-product semiring common in traditional linear algebra. We apply this semiring matrix factorisation framework on two tasks: sparse rank matrix factorisation and rank matrix tiling. Experiments on both synthetic and real world datasets show that the framework is capable of discovering different types of structure as well as obtaining high quality solutions
Stressors and satisfactors in entrepreneurial activity: an event-based, mixed methods study predicting small business owners' health
No full textInternational audienceEntrepreneurial activity leads to certain events that can impact both the equilibrium of a small business and that of its owner-manager. Although the health of the owner-manager is a key asset for a small organisation, the characteristics of the events he experiences and his psychological reaction to these episodes remain underexplored. This paper aims to address these research gaps, which are common to both occupational health and entrepreneurship, two fields that are rarely combined. We conducted a mixed methods study using a panel of 357 small business owners-managers. First, a qualitative longitudinal survey was administered to semantically categorise the most salient events experienced by the owners. Next, a quantitative, cross-sectional questionnaire was administered to weight the coded events through two mediating constructs: stress and satisfaction. Finally, we assessed the events' effects on the owners' health variances through linear regressions. Conceptually, our results extend the affective events theory to small business owners. Methodologically, we offer two checklists of events that capture emotional stress and satisfaction in entrepreneurial activity. Practically, certain preventive actions by medical professionals can build upon these checklists' predictive qualities regarding health. These checklists, referred to as a 'stressometer' and a 'satisfactometer', are also benchmarking tools for any self-employed
A Hybrid Robust Fault Tolerant Control Based on Adaptive Joint Unscented Kalman Filter
No full textInternational audienc
Adjustable Parameter-Based Multi-Objective Fault Estimation Observer Design for Continuous-Time/Discrete-Time Dynamic Systems
No full textInternational audienc
Improving Population Monte Carlo: Alternative Weighting and Resampling Schemes
No full textInternational audienc
Zonotope hit-and-run for efficient sampling from projection DPPs
No full textInternational audienceDeterminantal point processes (DPPs) are distributions over sets of items that model diversity using kernels. Their applications in machine learning include summary extraction and recommendation systems. Yet, the cost of sampling from a DPP is prohibitive in large-scale applications, which has triggered an effort towards efficient approximate samplers. We build a novel MCMC sampler that combines ideas from combinatorial geometry, linear programming, and Monte Carlo methods to sample from DPPs with a fixed sample cardinality, also called projection DPPs. Our sampler leverages the ability of the hit-and-run MCMC kernel to efficiently move across convex bodies. Previous theoretical results yield a fast mixing time of our chain when targeting a distribution that is close to a projection DPP, but not a DPP in general. Our empirical results demonstrate that this extends to sampling projection DPPs, i.e., our sampler is more sample-efficient than previous approaches which in turn translates to faster convergence when dealing with costly-to-evaluate functions, such as summary extraction in our experiments
Attitude towards Risk and Production decision : An Empirical analysis on French private forest owners.
No full textThis paper deals with the forest owner's attitude towards risk and the harvesting decision in several ways. First, we propose to characterize and quantify the forest owner's attitude towards risk. Second, we analyze the determinants of the forest owner's risk attitude. Finally, we determine the impact of the forest owner's risk attitude on the harvesting decision. The French forest owner's risk attitude is tackled by implementing a questionnaire, including a context-free measure borrowed from experimental economics. The determinants of the forest owner's risk attitudeand harvesting decision are estimated through a recursive bivariate ordered probit model. We show that French forest owners are characterized by a relative risk aversion coefficient close to 1 with a DARA assumption. In addition, we find that the forest owner's risk aversion is influenced positively and significantly by the level of risk exposure, the geographical location of the forest and the fact to be a forester, and negatively by the income. Finally, we obtain that the forest owner's risk aversion has a positive and significant impact on the harvesting decision
Subjective Fairness: Fairness is in the eye of the beholder
No full textWe introduce a natural, and widely applicable framework for fairnessthat relies on the available information. We develop algorithms for achievinga few different notions of fairness within a subjective framework, and in particularrecently proposed concepts of fairness that are grounded in concepts ofsimilarity and conditional independence. We argue that a suitable notion ofsimilarity in the Bayesian setting is distributional similarity conditioned on theobservations. For the latter, as independence is difficult to achieve uniformlyin the Bayesian setting, we suggest a relaxation, for which we provide a smallexperimental demonstration
Language Support for Modular Autonomic Managers in Reconfigurable Software Components
No full textInternational audienceDynamic reconfiguration is a key capability of Component-based Software Systems to achieve self-adaptation as it provides means to cope with environment changes at runtime. The space of configurations is defined by the possible assemblies of components, and navigating this space while achieving goals and maintaining structural properties is managed in an auto-nomic loop. The natural architectural structure of component-based systems calls for hierarchy and modularity in the design and implementation of composites and their managers, and requires support for coordinated multiple autonomic loops. In this paper, we leverage the modularity capability to strengthen the Domain-Specific Language (DSL) Ctrl-F, targeted at the design of autonomic managers in component-based systems. Its original definition involved discrete control-theoretical management of reconfigurations, providing assurances on the automated behaviors. The objective of modularity is twofold: from the design perspective, it allows designers to seamlessly decompose a complex system into smaller pieces of reusable architectural elements and adaptive behaviours. From the compilation point of view, we provide a systematical and generative approach to decompose control problems described in the architectural level while relying on mechanisms of modular Discrete Control Synthesis (DCS), which allows us to cope with the combinatorial complexity that is inherent to DCS problems. We show the applicability of our approach by applying it to the self-adaptive case study of the existing RUBiS/Brownout eBay-like web auction system
Comment devenir Trajan quand on n'est qu'Honorius ?
No full textInternational audienceTrajan est regardé comme l’un des modèles, si ce n’est le modèle par excellence, de l’optimus princeps dans l’Antiquité tardive. On retrouve ainsi plusieurs références implicites ou explicites à Trajan dans les panégyriques impériaux de Claudien. Faut-il en conclure qu’il use de cette référence comme d’un simple topos encomiastique, qui concourt à décrire Honorius comme le nouvel optimus princeps de l’Empire ?Quand Claudien reprend à son compte l’idée de la supposée filiation entre Trajan et Théodose, on remarque qu’au lieu de s’en servir pour rehausser le prestige du jeune empereur, il souligne plutôt l’écart qui existe entre lui et son glorieux aïeul. Si Théodose s’est montré digne de son appartenance à la gens Ulpia et a obtenu le pouvoir, c’est par ses exploits militaires et sa vertu, non par l’intrigue ou par la naissance. Ainsi, dans le cas de son fils, une naissance prestigieuse ne peut suffire à lui assurer la légitimité à la tête de l’Empire d’Occident et les louanges du poète. On voit ici une tension entre la transmission dynastique du pouvoir et l'idée selon laquelle on doit avoir à la tête de l'Empire l’homme le meilleur qui soit. Néanmoins, le poète lui montre aussi la voie à suivre. L’image de l’optimus princeps et de l’empereur-citoyen, que nous transmet la mémoire de Trajan via Pline et les Panégyriques latins, est aussi un idéal à atteindre. Honorius ne mérite pas d’être loué car il est aussi bon empereur que l’a été Trajan ; mais il méritera de l’être si/quand il suivra le programme prescrit par le poète. À ce titre, le panégyrique n’est plus tant louange que contrat passé entre l’encomiaste (qui se fait conseiller du Prince) et son destinataire