Paris Dauphine University
Recent publications
We study the problem of selecting a single element from a set of candidates on which a group of agents has some spatial preferences. The exact distances between agent and candidate locations are unknown but we know how agents rank the candidates from the closest to the farthest. Whether it is desirable or undesirable, the winning candidate should either minimize or maximize its aggregate distance to the agents. The goal is to understand the optimal distortion, which evaluates how good an algorithm that determines the winner based only on the agent rankings performs against the optimal solution. We give a characterization of the distortion in the case of latent Euclidean distances such that the candidates are aligned, but the agent locations are not constrained. This setting generalizes the well-studied setting where both agents and candidates are located on the real line. Our bounds on the distortion are expressed with a parameter which relates, for every agent, the distance to her best candidate to the distance to any other alternative.
The long-time average behavior of the value function in the calculus of variations is known to be connected to the existence of the limit of the corresponding Abel means. Still in the Tonelli case, such a limit is in turn related to the existence of solutions of the critical (or ergodic) Hamilton-Jacobi equation. The goal of this paper is to address similar issues when set on the whole Euclidean space and the Hamiltonian fails to be Tonelli. We first study the convergence of the time-averaged value function as the time horizon goes to infinity, proving the existence of the critical constant (Mañé critical value) for a general control system. Then, we show that the ergodic equation admits solutions for systems associated with a family of vector fields which satisfies the Lie Algebra rank condition. Finally, we construct a solution to the critical HJB equation on the whole space which coincides with its Lax-Oleinik evolution.
The Central Andes of Peru are a region of great concern regarding pesticide risk to the health of local communities. Therefore, we conducted an observational study to assess the level of pesticide contamination among Andean people. Analytical chemistry methods were used to measure the concentrations of 170 pesticide-related compounds in hair samples from 50 adult Andean subjects living in rural and urban areas. As part of the study, a questionnaire was administered to the subjects to collect information regarding factors that increase the risk of pesticide exposure. Our results indicate that Andean people are strongly exposed to agrochemicals, being contaminated with a wide array of pesticide-related compounds at high concentration levels. Multivariate analyses and geostatistical modeling identified sociodemographic factors associated with rurality and food origin that increase pesticide exposure risk. The present study represents the first comprehensive investigation of pesticide-related compounds detected in body samples collected from people living in the Central Andes of Peru. Our findings pinpoint an alarming environmental situation that threatens human health in the region and provide a rationale for improving public policies to protect local communities.
From the US to Italy, from Brazil to Japan, cities from all over the world are increasingly vocal on migration issues. Advocating for alternative approach to immigrants' welcome, their stand and policies may at times be in blunt contradiction with national approaches. This paper gives an overview of this new form of urban militancy, its recent evolution, its forms, its networks. Drawing on case studies in France, Spain and Italy, it seeks to explain why the Mediterranean has been an important setting for the politicization of municipal involvement and the rise of multi-actor power assemblage of municipalities and voluntary organizations.
State Medical Aid is a public health insurance program that allows undocumented immigrants with low financial resources to access health care services for free. However, the low take-up rate of this program might threaten its efficiency. The purpose of this study is therefore to provide the determinants of such a low take-up rate. To this end, we rely on the Premier Pas survey. This is an original representative sample of undocumented immigrants attending places of assistance to vulnerable populations in France. Determinants of State Medical Aid take-up are analyzed through probit and Cox modeling. The results show that only 51% of those who are eligible for the State Medical Aid program are actually covered, and this proportion is higher among women than among men. The length of stay in France is the most important determinant of take-up. It is worth noting that State Medical Aid take-up is not associated with chronic diseases or functional limitations and is negatively associated with poor mental health. There is, therefore, mixed evidence of health selection into the program. Informational barriers and vulnerabilities experienced by undocumented immigrants are likely to explain this low take-up.
Introduction An increase in migration rates to the European Union has been observed over the last few years. Part of these migrants is undocumented. This work aimed to describe the reported frequency of infectious diseases and their associated factors among unselected samples of undocumented migrants in France. Methodology The Premier Pas survey is a cross-sectional epidemiological survey of a random sample (two-stage sample design) conducted among undocumented migrants recruited in Paris and the Bordeaux region, in places and facilities likely to be frequented by undocumented migrants. The percentages were weighted. The analysis was performed using Stata 15.1 software. Results A total of 1,223 undocumented migrants were recruited from 63 places and facilities, with a participation rate of 50%. Most of them were between 30 and 40 years of age (36%), 69% were men, aged mainly 30–40 (36%) years old, from sub-Saharan Africa (60%) or North Africa (25%), and 60% had arrived <3 years earlier. Among the participants, 24.8% declared a poor perceived health status and 33.5% a chronic health condition. Dental infections concerned 43.2% of the participants. Apart from dental issues, 12.9% reported suffering from at least one infectious disease: HIV infection (3.5%), chronic hepatitis B virus infection (3.1%), upper respiratory tract infection (1.7%), skin mycosis (1.2%), skin and soft tissue infection (0.8%), chronic hepatitis C infection (0.8%), urinary tract infection (0.7%), lower respiratory tract infection (0.7%), scabies (0.3%), tuberculosis disease (0.2%), vaginal mycosis (0.6%), and herpes (0.1%). Regarding HIV, HBV, and HCV infections, 56, 71, and 89%, respectively, were diagnosed after their arrival. Chronic viral infections were more often reported by undocumented migrants from sub-Saharan Africa and Latin America. In multivariate analysis, a higher risk of reporting chronic viral infection was observed among people food insecure. Conclusion This original study on a large random sample confirms the frequency of infectious diseases among undocumented migrants in France and the importance of integrating their screening during a health Rendezvous and their management into early access to care and inclusive medico-psycho-social management.
Diapeutics gene markers in colorectal cancer (CRC) can help manage mortality caused by the disease. We applied a game-theoretic link relevance Index (LRI) scoring on the high-throughput whole-genome transcriptome dataset to identify salient genes in CRC and obtained 126 salient genes with LRI score greater than zero. The biomarkers database lacks preliminary information on the salient genes as biomarkers for all the available cancer cell types. The salient genes revealed eleven, one and six overrepresentations for major Biological Processes, Molecular Function, and Cellular components. However, no enrichment with respect to chromosome location was found for the salient genes. Significantly high enrichments were observed for several KEGG, Reactome and PPI terms. The survival analysis of top protein-coding salient genes exhibited superior prognostic characteristics for CRC. MIR143HG, AMOTL1, ACTG2 and other salient genes lack sufficient information regarding their etiological role in CRC. Further investigation in LRI methodology and salient genes to augment the existing knowledge base may create new milestones in CRC diapeutics.
This paper investigates the theory of robustness against adversarial attacks. We focus on randomized classifiers (i.e. classifiers that output random variables) and provide a thorough analysis of their behavior through the lens of statistical learning theory and information theory. To this aim, we introduce a new notion of robustness for randomized classifiers, enforcing local Lipschitzness using probability metrics. Equipped with this definition, we make two new contributions. The first one consists in devising a new upper bound on the adversarial generalization gap of randomized classifiers. More precisely, we devise bounds on the generalization gap and the adversarial gap i.e. the gap between the risk and the worst-case risk under attack) of randomized classifiers. The second contribution presents a yet simple but efficient noise injection method to design robust randomized classifiers. We show that our results are applicable to a wide range of machine learning models under mild hypotheses. We further corroborate our findings with experimental results using deep neural networks on standard image datasets, namely CIFAR-10 and CIFAR-100. On these tasks, we manage to design robust models that simultaneously achieve state-of-the-art accuracy (over 0.82 clean accuracy on CIFAR-10) and enjoy guaranteed robust accuracy bounds (0.45 against ℓ2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\ell _{2}$$\end{document} adversaries with magnitude 0.5 on CIFAR-10).
In this paper we prove a regularity and rigidity result for displacements in $$GSBD^p$$ G S B D p , for every $$p>1$$ p > 1 and any dimension $$n\ge 2$$ n ≥ 2 . We show that a displacement in $$GSBD^p$$ G S B D p with a small jump set coincides with a $$W^{1,p}$$ W 1 , p function, up to a small set whose perimeter and volume are controlled by the size of the jump. This generalises to higher dimension a result of Conti, Focardi and Iurlano. A consequence of this is that such displacements satisfy, up to a small set, Poincaré-Korn and Korn inequalities. As an application, we deduce an approximation result which implies the existence of the approximate gradient for displacements in $$GSBD^p$$ G S B D p .
A voting rule is monotonic if a winning candidate never becomes a loser by being raised in voters’ rankings of candidates, ceteris paribus. Plurality with a runoff is known to fail monotonicity. To see how widespread this failure is, we focus on French presidential elections since 1965. We identify mathematical conditions that allow a logically conceivable scenario of vote shifts between candidates that may lead to a monotonicity violation. We show that eight among the ten elections held since 1965 (those in 1965 and 1974 being the exceptions) exhibit this theoretical vulnerability. To be sure, the conceived scenario of vote shifts that enables a monotonicity violation may not be plausible under the political context of the considered election. Thus, we analyze the political landscape of these eight elections and argue that for two of them (2002 and 2007 elections), the monotonicity violation scenario was plausible within the conjuncture of the time.
Standards are a core part of risk regulation, acting as one of the key resources for technical professionals and their companies in the design and long-term management of hazardous infrastructure. Yet they lurk in the background, in so far as safety scientists rarely attend to what makes a particular standard effective. This paper contributes to the literature by focusing on the social aspects of standard formation and use, taking as our case a recent update to one part of AS 2885, the Australian Standard for pipeline engineering. Working with the ideas of ‘input’ and ‘output’ legitimacy in the eyes of end users, we address justification of the national standard, committee membership and the interests it reflects, and views about standard effectiveness. We identify high levels of input legitimacy for the standard due to its institutional links and high technical expertise of subcommittee members. The effectiveness of the standard to date confers output legitimacy. Input and output legitimacy, together, underlie standard compliance. The case suggests that compliance with safety standards more broadly might be supported by a higher profile for standards writers themselves. The link between industry culture and standards is emphasised as is the value of involving regulators in standard development.
Despite recent progress, about 295,000 women in the World still die each year from pregnancy-related causes, and about 4.1 million children die before reaching the age of one. 99% of these deaths occur in developing countries. In 2006 the Zambian government removed user fees in public and mission health facilities in 54 out of 72 districts, and then extended this policy to rural parts of unaffected districts in 2007. I exploit the staggered implementation of the policy to assess its impact on maternal health care utilization and child health outcomes. Using a difference-in-differences estimation strategy, I find a 43% increase in the probability to give birth in a medical facility following the removal and a 36% increase in the probability of being assisted by a skilled birth attendant during childbirth. These positive effects decrease with household’s distance from the nearest health facility. In terms of child health, chronic malnutrition decreased by 8% and the abolition of user fees reduced newborn mortality risk only for those living close to a health facility providing essential emergency obstetric care and child health services. Access improved but returns to formal health services remained rather limited, highlighting the importance of addressing supply-side constraints to generate substantial gains in population health.
We study the one-dimensional asymmetric simple exclusion process on the lattice {1,⋯,N}\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\{1, \dots ,N\}$$\end{document} with creation/annihilation at the boundaries. The boundary rates are time dependent and change on a slow time scale N-a\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N^{-a}$$\end{document} with a>0\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$a>0$$\end{document}. We prove that at the time scale N1+a\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$N^{1+a}$$\end{document} the system evolves quasi-statically with a macroscopic density profile given by the entropy solution of the stationary Burgers equation with boundary densities changing in time, determined by the corresponding microscopic boundary rates. We consider two different types of boundary rates: the “Liggett boundaries” that correspond to the projection of the infinite dynamics, and the reversible boundaries, that correspond to the contact with particle reservoirs in equilibrium. The proof is based on the control of the Lax boundary entropy–entropy flux pairs and a coupling argument.
The architecture of neural networks in neural based computer game programs influences greatly the strength of the game playing programs. We present developments on the recently tested Mobile Network architecture that has good results for the game of Go. The three proposed improvements deal with the optimization process, the activation function and the convolution layers. These three modifications improve the accuracy of the policy and the error of the evaluation, as well as the playing strength of a computer Go program using the resulting networks.
The efficiency of Monte-Carlo based algorithms heavily relies on a random search heuristic, which is often hand-crafted using domain knowledge. To improve the generality of these approaches, new algorithms such as Nested Rollout Policy Adaptation (NRPA), have replaced the hand crafted heuristic with one that is trained online, using data collected during the search. Despite the limited expressiveness of the policy model, NRPA is able to outperform traditional Monte-Carlo algorithms (i.e. without learning) on various games including Morpion Solitaire. In this paper, we combine Monte-Carlo search with a more expressive, non-linear policy model, based on a neural network trained beforehand. We then demonstrate how to use this network in order to obtain state-of-the-art results with this new technique on the game of Morpion Solitaire. We also use NeuralNRPA as an expert to train a model with Expert Iteration.
Risk is a complex strategy game that may be easier to understand for humans than chess but harder to deal with for computers. The main reasons are the stochastic nature of battles and the different decisions that must be coordinated within turns. Our goal is to create an artificial intelligence able to play the game without human knowledge using the Expert Iteration [1] framework. We use graph neural networks [13, 15, 22, 30] to learn the policies for the different decisions and the value estimation. Experiments on a synthetic board show that with this framework the model can rapidly learn a good country drafting policy, while the main game phases remain a challenge.
Institution pages aggregate content on ResearchGate related to an institution. The members listed on this page have self-identified as being affiliated with this institution. Publications listed on this page were identified by our algorithms as relating to this institution. This page was not created or approved by the institution. If you represent an institution and have questions about these pages or wish to report inaccurate content, you can contact us here.
3,112 members
Thierry Kirat
  • Center for Interdisciplinary Research in the Social (IRISSO)
Choukri Hmed
  • Center for Interdisciplinary Research in the Social (IRISSO)
Béatrice Parguel
  • Dauphine Center for Management Research (DRM)
jean-François Chanlat
  • Management and Organization DRM UMR-CNRS 7088
Gabriella Pigozzi
  • Laboratory for Analysis and Modeling Systems for Decision Support (LAMSADE)
Place du Maréchal de Lattre de Tassigny, 75775, Paris, France