The main objective of the project is to propose a novel approach of distributed, scalable, dynamic and energy efficient algorithms for managing resources in a mobile network. This new approach relies on the design of an orchestration mechanism of a portfolio of algorithms. The ultimate goal of the proposed mechanism is to enhance the user experience, while at the same time to better utilize the operator resources. User mobility and new services are key elements to take into account if the operator wants to improve the user quality of experience. Future autonomous network management and control algorithms will thus have to deal with a real-time dynamicity due to user mobility and to traffic variations resulting from various usages. To achieve this goal, we focus on two central aspects of mobile networks and intend to design distributed learning mechanisms in non-stationary environments, as well as an orchestration mechanism that applies the best algorithms depending on the situation. The first main aspect to be addressed is the management of radio resources at the RAN (Radio Access Network) level. In current (LTE) and future (LTE-A) cellular systems, interference appears as a bottleneck for providing high data rates and seamless connectivity to the end-user. To reduce interference it is possible either to coordinate the transmissions of neighboring base stations (BS) so as to avoid simultaneous transmissions on the same radio resources or to allow BSs to cooperate: two or more BSs combine their transmissions towards a single user in order to increase its data rate. Both cases require distributed learning algorithms. The second aspect is the management of the popular contents users want to get access to. In a Content Delivery Network (CDN), popular content is disseminated and stored in cache servers as close as possible to the demand to avoid delay in access. How to place servers in the network and replicated contents in the servers are traditional issues in CDNs. In mobile CDNs, things are exacerbated because of the changing and unpredictable environment characterized by spatial and temporal changes in the traffic demand, user mobility and variable channel conditions. The way the project intends to tackle these problems is based on a “learn to learn approach”. If we think about BSs and cache servers as autonomous entities seeking to optimize a global objective function and able to take decisions based on incomplete information, the notion of distributed learning arises naturally. There are numerous approaches along these lines and each mechanism has its own characteristics in terms of needed information, type of achieved equilibrium, convergence speed, and stability. Each mechanism can also be tuned thanks via an array of parameters. The problem is exacerbated in non-stationary situations due to mobility, traffic demand or radio channel variations. The originality of the project thus lies in its objective of building a portfolio of distributed learning algorithms that are then to be orchestrated. To account for learning in the presence of non-stationary processes, we intend to use the theory of stochastic approximation in order to develop robust versions of existing learning schemes. Orchestrating a portfolio of learning algorithms is, in many regards, similar to the literature on “learning with expert advice”, so our goal will be to devise adaptive learning schemes that select dynamically between different learning schemes so that their long-term learning power exceeds the regret of any individual “expert”. Bringing together experts form both network and learning, NETLEARN ultimately intends to propose architecture and protocol adaptations for implementing our resource management algorithms.

Recent advances of statistics have allowed for developments of powerful methodologies to analyze data from both the theoretical and the methodological point of view. However, the issue of calibrating the parameters involved in modern statistical procedures remains most of the time an open question, raising many difficulties for practitioners. This ANR proposal, entitled "statistical calibration" aims at providing some remedies to fulfill this gap and strengthening collaborations between theoreticians and practitioners. Even if for obvious reasons, calibration aspects have always been considered, at least in practice, the calibration of statistical procedures constitutes a fundamental, challenging and topical axis of research. We intend to deal with this question by using modern mathematical technics such as non-asymptotic concentration inequalities, the contribution of the Bayesian approach and the recent developments of classical methods such as $V$-fold cross-validation, bootstrap and resampling or more innovative ones such as random forests. We aim at facing with various problems such as supervised and non-supervised classification, non-parametric estimation, detection, testing... To address these problems, all the modern methodologies will be considered. The structure of our proposal detailed in the scientific file emphasizes the importance of practical aspects in our project (several non-mathematician experts are involved). Even if theoretical results will be investigated, this choice highlights that concrete aspects of calibration and the use of our methods by practitioners will be our main motivation. Specialists of calibration and of the topics mentioned previously are gathered in the ANR "Calibration" proposal. More precisely, this project is composed of 34 members (27 have a permanent position, and 7 are PhD students or postdocs). Most of them have a position in one of the three partners involved in the project : CEREMADE (Paris Dauphine), Laboratoire Jean-Alexandre Dieudonné (University of Nice) and Laboratoire de Mathématiques of the University Paris Sud. The coordinator of the project is Vincent Rivoirard (University Paris Dauphine). Scientific tasks are coordinated by Pascal Massart (University Paris Sud - Orsay), Sylvain Arlot (CNRS and ENS Paris) and Patricia Reynaud-Bouret (CNRS and University of Nice). The requested budget for this proposal is 590 148 euros. These funds will be allocated to finance three post-doctorate positions of one year (one for each partner), missions, invitations of researchers, computer equipments, working meetings and one international workshop on the topic of statistical calibration.

Social sciences of work and politics have poorly documented the relationship between health and political activity. However, research focusing on elected representatives suggests that there is a need to study this subject. On the one hand, the literature emphasises the intensity of political work, on the other hand, it reminds us that dedication is a central component of the political ethos. This tension, which is structurally inscribed in political activity, invites us to consider the health of elected representatives as an object of research. To do so, we hypothesise that tensions between, on the one hand, multiple forms of testing and wear and tear resulting of the requirements of the function, which can potentially degrade health, and, on the other hand, injunctions to dedication and norms of conduct requiring good health, affect the exercise of political mandate(s). The ELUSAN project, which focuses on professional elected officials (national and local), will contribute to enriching and renewing knowledge of the political profession, by combining contributions from the sociology of work and political science. The objective is to answer four linked questions: What are the salient features of the working conditions of elected representatives? How do tacit professional norms on health circulate in the political field? How has the institutional protection of elected officials' health been differentiated and unequal? How is health inscribed in work experiences and political careers? Finally, the ELUSAN project seeks to make a double break. A break with ordinary but also indigenous discourses that tend to deny any physical or psychological weaknesses to elected representatives and a break with academic approaches to politics that do not consider health as a significant component of political activities.

Since 1975, when they were only 76 in 25 countries, the number of « export processing zones » (EPZ) has reached 3500 in 130 countries in 2006. The number of workers in these zones is assumed to have tripled. According to the OECD, free trade zones “constitute a second best…, because they benefit only a small number and create distortions in resource allocation, but they could encourage exchange liberalization a the country level”. The goal of this project is to study more precisely this conclusion, and, if necessary, criticizes it, by better identifying the effects of free zones. The starting point of this project is a transversal study of export processing zones, before going into more details with a study of the nature and impact of the Manaus Free Trade Zone (MFZ), created in 1967 by the Brazilian government at the heart of the Amazonian forest, on topics as trade, poverty, inequalities, and natural, economic and social environment. Despite their growing importance, free zones have indeed been scarcely studied at the empirical as well as at the theoretical level. While the Asian free trade zones, in particular the Chinese zones, are better known, the MFZ has never been systematically studied. Free trade zones do not have their own theory, in the same way that regional integration does, for example. The transversal study will show where the MFZ stands among other free zones. It will include descriptive, theoretical and empirical aspects (using the gravity model) and will shed light on the debate over the effects of these zones on trade. Free zones “distort” the nature of specialization itself, less from the point of view of the final products than centred on the steps of the production process itself in the “vertical” division of labour framework. A very precise study of a micro-territory –the MFZ-, should allow a better quantification of the extent of this phenomenon, which has not been properly assessed yet, by centring the analysis on the fragmentation of the added value in the zone. Studies of the influence of trade openness - accelerated and configured by free zones- on inequalities and poverty have often concluded that there is a negative relationship between openness and poverty and a positive one between openness and inequalities. The MFZ represents a particularly interesting field because of its very high degree of trade openness centred on importations which allows us to better isolate the impact of openness. Social practices of free zones have often been challenged. Are free zones leading other countries (and here, other regions of the same country) in a race-to-the-bottom process by using fiscal, social or environmental “dumping” mechanisms? Keeping in mind the history and traditions of the Manaus region, a detailed analysis of the effects of the free zone is necessary with respect to labour laws, women’s situation, consequences on the indigenous people hired in the zone, environmental sensitivity of the region, in particular to transportation, water management and energetic infrastructures. A field survey will be carried out in Manaus and the surrounding region to answer these questions. These various studies rely on a theoretical framework that isolates contradictory effects of free zones and will be based on advanced econometric methods using the available data at a fine level or survey data collected for this purpose. The project should lead to refine results on the effect of free zones in general and the MFZ in particular: effects on internal and external trade, on industrial and regional development, poverty, inequalities and social practice of implemented firms in the zone. These results should shed light on the benefits and drawbacks of export processing zones for Brazil and beyond that, bring information to the stakeholders (NGOs, trade unions, international organizations, and trade policymakers) on the appropriateness of the exceptional status of free zones.