Network diffusion processes -- that is, the dynamics by which some information, behavior, or condition starting out with a single or a few individuals, propagates through the network to reach a potentially much larger number -- are ubiquitous in social, biological, and technological networks. The spread of rumors, trends and innovation in social networks, the spread of diseases and mutations in population networks, or viruses and worms in computer networks are just a few examples. In this project, we will study the effect of the network structure on diffusion processes, and in particular, the extent to which one can affect the outcome of the diffusion by controlling a small, possibly, carefully selected fraction of the network. For example, increase the spread or speed of diffusion by choosing an appropriate set of seed nodes (a standard goal in viral marketing by word-of-mouth), or achieve the opposite effect either by choosing a small set of nodes to remove (a goal in immunization against diseases), or by seeding a competing diffusion (e.g., to limit the spread of misinformation in a social network). Our goal is to provide a framework for a systematic and rigorous study of these problems. We will consider several standard diffusion models and extensions of them, including models from mathematical sociology, mathematical epidemiology, and interacting particle systems. We will consider existing and new variants of spread maximization/limitation problems, and will provide (approximation) algorithms or show negative (inapproximability) results. In case of negative results, we will investigate general conditions that make the problem tractable. We will consider both general network topologies and specific network models, and will relate the efficiency of solutions to structural properties of the topology. Finally, we will use these insights to engineer new network diffusion processes for efficient data dissemination.

Project at a glance Crowdsourcing relies on potentially huge numbers of on-line participants to resolve data acquisition or analysis tasks. It is an exploding area that impacts various domains, ranging from scientific knowledge enrichment to market analysis support. But currently, existing crowd platforms rely mostly on low level programming paradigms, rigid data models and poor participant profiles, which yields severe limitations. The low-level nature of existing solutions prevents the design of complex data acquisition workflows, that could be executed, composed, searched and even be proposed by participants themselves. Taking into account the quality, uncertainty, inconsistency and representativeness of participant contributions is still an open problem. Methods for assigning a task to the correct participant according to his trust, motivation and expertise, automatically improving crowd execution time, computing optimal participant rewards, are missing. Similarly, usual crowd campaigns produce isolated and rigid data sets: A flexible and common data model for the produced knowledge about data and participants could allow participative knowledge acquisition. To overcome these challenges, Headwork will define: - Rich workflow, participant, data and knowledge models to capture various kind of crowd applications with complex data acquisition tasks and human specificities, - Methods for deploying, verifying, optimizing, but also monitoring and adapting crowd-based workflow executions at run time. To reach this goal, Headwork will rely on two experts of large participative knowledge acquisition platforms MNHN-Cesco & FouleFactory, from academy and industry respectively), on major academic teams on data management and workflow modeling (Dahu, Druid,Links, Sumo)... and on a crowd of around 60,000 registered participants from our platforms.

This project is aligned with the theme 1 "Sécurité et sûreté des systèmes numériques". Considering that reverse engineering of code and the recovery of sensitive data are nowadays one of the greatest threats in secure devices, COGITO proposes an innovating solution that puts into perspective the use of runtime code generation in order to increase the level of security in embedded information systems. Security in embedded devices and runtime code generation are, a priori, two technological fields that hardly combine together. On one hand, secure elements must target small production costs, silicon and energy consumption, and as such, offer very limited computing and memory resources. On the other hand, compilation is a computation-intensive process, and dynamic compilation techniques require a fair amount of computing power and of memory resources at runtime. However, the objective of the COGITO project is to demonstrate the applicability and the effectiveness of code generation techniques applied at runtime and on board for security purposes in embedded devices. In this project we will define and validate a unique protection mechanism that implements a wide range of ad hoc countermeasures and provide a means for effective code obfuscation. The objective of the “factorization” of a large set of countermeasures is to obtain a better trade-off between security and performance than the state-of-the art solutions. To reach this objective, the partners of the project COGITO plan to adapt a technology for runtime code generation developed, by the CEA. This technology, called deGoal, is fundamentally different from the traditional approaches (interpretation and dynamic compilation): ad hoc code generators compiled statically and are embedded in the target application, each code generator being dedicated for each computing kernel whose binary code will be updated at runtime. Thus, these code generators are lightweight and very fast, allowing to target small architectures that are usually out of reach of the standard techniques for dynamic code generation such as the small microcontrollers used in secure devices. Furthermore, we are confident about the ability of our solution to combine well with other software and hardware state-of-the-art countermeasures for cryptography. The three main tasks to achieve the objective are : 1. Provide an in-depth analysis of the opportunities and threats of runtime code generation to increase the level of security in secure devices. 2. Demonstrate the applicability of runtime code generation to the field of secure devices. To achieve this objective, we will adapt the tool deGoal. 3. Experiment, measure and validate the effectiveness of runtime code generation in illustrative use cases. In parallel to these technical tasks, the dissemination of the project results will be carried out via a dedicated website, via publications in outstanding journals in the fields of interest, via the participation in conferences, workshops and the events organised by the ANR, and via the organization of a special workshop at mid term focusing industrials in particular.

The past decade has witnessed a tremendous interest in the concept of sparse representations in signal and image processing. One of the main reasons explaining this enthusiasm stands in the discovery of compressive sensing, a new sampling paradigm defying the theoretical limits established sixty years before by Shannon. Compressive sensing led many researchers to focus on inverse problems involving fairly-well conditioned dictionaries as those arising from random, independent measurements. Yet in many applications, the dictionaries relating the observations to the sought sparse signal are deterministic and ill-conditioned. In these scenarios, many classical algorithms and theoretical analyses are likely to fail. The BECOSE project aims to extend the scope of sparsity techniques much beyond the academic setting of random and well-conditioned dictionaries. 1. Conception of new algorithms Inverse problems exploiting the sparse nature of the solution rely on the minimization of the counting function, referred to as the L0-"norm". This problem being intractable in most practical settings, many suboptimal resolutions have been suggested. The conception of algorithms dedicated to ill-conditioned inverse problems will revolve around three lines of thought. First, we will step back from the popular L1-convexification of the sparse representation problem and consider more involved nonconvex formulations. Recent works indeed demonstrate their relevance for difficult inverse problems. However, designing effective and computationally efficient algorithms remains a challenge for problems of large dimension. Second, we will study the benefit of working with continuous dictionaries in contrast with the classical discrete approach. Third, we will investigate the exploitation of additional sources of structural information (on top of sparsity) such as non-negativity constraints. 2. Theoretical analysis of algorithms The theoretical analysis aims at characterizing the performance of heuristic sparse algorithms. The traditional worst-case exact recovery guarantees are acknowledged to be rather pessimistic because they may not reflect the average behavior of algorithms. It is noticeable, though, that sharp worst-case exact recovery conditions are not even available for a number of popular L0 algorithms. We will focus on stepwise orthogonal greedy search algorithms, which are very well-suited to the ill-conditioned context. We foresee that they will enjoy much weaker recovery guarantees than simpler L0 algorithms. We further propose to elaborate an average analysis of greedy algorithms for deterministic dictionaries, which is a major open issue. To do so, several intermediate steps will be carried out including a guaranteed failure analysis and the derivation of weakened guarantees of success by taking into account other constraints on top of sparsity such as prior knowledge on the signs, coefficient values, and partial support information. 3. From theory to practice The proposed algorithms will be assessed in the context of tomographic Particle Image Velocimetry (PIV), a rapidly growing imaging technique in fluid mechanics that will have strong impact in several industrial sectors including environment, automotive and aeronautical industries. This flow measurement technique aims to determine the 3D displacement of tracer particles that passively follow the flow, based on the acquisition of a limited number of 2D camera images. The resulting inverse problem involves high-dimensional data as a time sequence of highly resolved 3D volumes must be reconstructed. Presently available methods for 3D reconstruction and flow tracking are still restricted to small volumes, which is the main bottleneck together with accuracy and resolution limits. The sparse approach is the key methodological tool to handle problems of larger dimension. The proposed solutions will be validated using both realistic simulators and real experimental data.

The objective of this project, called GDiv, is to setup a strong network of European partners around the core team composed of INRIA (coordinator) and SINTEF. This network will gather another academic partner and between 3 and 5 industry partners in the areas of software development and deployment. This network will be setup in order to prepare a project proposal for the call ICT10-2016 Software Technologies. I am currently coordinating a FET project (FP7 - DIVERSIFY), which brings together researchers from the domains of software-intensive distributed systems and ecology in order to translate ecological concepts and processes into software design principles. In particular, we investigate how the dynamics of biodiversity can be adapted to increase the diversity in software systems in order to increase their robustness against unpredictable environmental perturbations (bugs, hardware failures, attacks, network latency, etc.). This project has developed fundamental principles that are now ready for transfer into actual new software technology for the engineering of safe large-scale software systems. The main goal of GDiv is to setup a consortium, which gathers partners from industry and academia in the area of software technologies and that fits the LEIT-ICT spirit. From a research and innovation perspective, the project proposal setup by the GDiv network will address the risks of large scale software reuse through integrated, multi-level software diversification techniques. Software reuse is essential to build the large-scale software applications that pervade our daily lives. Yet, massive reuse has a darker side: it creates a monoculture of software applications and millions of clone programs around the world can be hacked in the same way. For example, Wordpress web sites have been the targets of an agressive hacking campaign in France, in the aftermath of the Charlie Hebdo attacks. Two factors favored this massive attack: (i) Wordpress is the dominating technology to build web sites (forming an applicative monoculture) and (ii) Wordpress developers introduced some rigidity in the code (e.g., ``hard-coded'' naming conventions), which favored the reconnaissance phase of these attacks. The project setup by the Gdiv network will aim at designing new software technologies to (i) automatically create large quantities of program variants that all provide similar functionality but implement diverse computation and (ii) integrate these variants in the deployment and maintenance processes. This kind of software diversity aims at reducing the risks of applicative software monoculture, while letting the developers benefit from code reuse.