he LeMOn project is an academic project with strong potential repercussions in many application domains. It takes place at the frontiers between three research communities that strongly interplay : machine learning, pattern recognition and multi-objective optimization. The rationale behind this project stems from the fact that most existing learning machines and pattern recognition systems are designed by optimizing a single criterion for a single task, whereas multiple criteria and/or multiple tasks are generally involved in real-world applications. As examples, one can cite the trade-offs between learning performance and generalization ability, between sensitivity and specificity or between the scalability and the decision speed. Hence, machine learning is inherently a multi-objective optimization problem and the aim of the LeMOn project is to initiate a break in the theory and the methodology of machine learning, by taking into account multiple criteria and tasks in the learning machine design. In this general framework, the LeMOn project propose to tackle two particular multi-objective learning problems : (1) The first problem concerns learning in undefined or weakly defined contexts, i.e. when a priori probabilities and/or misclassification costs are unknown. In such a context, it is well known that a single learning criterion does not lead to a classifier able to face to many sitations. In a previous work, we have proposed a multi-model selection framework based on a Pareto-based multi-objective approach working in the ROC space. By introducing the "ROC front concept", the idea was to train a pool of classifiers instead of a single one, each classifier in the pool optimizing a particular trade-off between the objectives. In the LeMOn project, we plan to extend this approach to large-scale problems, by considering on-line learning and to multi-class problems by investigating recent advances in multi-class ROC analysis. (2) The second problem concerns the Multi-Task Learning framework and its strong links with multi-objective optimization. Multi-Task Learning (MTL) is a statistical learning framework which seeks at learning several models in a joint manner. The idea behind this paradigm is that, when the tasks to be learned are similar enough or are related in some sense, it may be advantageous to take into account these relations between tasks. Such a framework strongly relies on an multi-objective optimization process where compromise have to be done between objectives related to each task. In the LeMOn project, we will focus on two speficic problems occuring in a MTL problem tackled with kernel methods. The first one consists in optimizing the choice of mixed norm penalty in the regularizing term while the second aims at finding the best weighting of each task. In each case, the proposed algorithms will be mainly evaluated on two particular applications which are inherently multi-objective and multi-task : Brain Computer Interfaces and medical image classification. Other applications such as document image analysis or fraud detection may also be envisaged.

This project aims at re-evaluating the synthetic potential of nitroarenes compounds. Most of these compounds are easily accessible and produced on large scales by a classical aromatic electrophilic nitration reaction. However, up to now, their synthetic use is, up to now, essentially devoted to their reduction into aromatic amines. Our goals are thus to develop the chemistry of nitroarenes and enlarge the panel of applicable transformations to increase their synthetic potential in view of accessing aminated compounds of high added value. We propose to develop new reactions involving the dearomatization of nitrobenzenes, via [4+2] cycloaddition reactions. In this inverse electron demand process, we expect the nitroarene to behave as an electron-poor heterodiene (C=C-N=O) by involving one of its aromatic C=C bonds in a cycloaddition with electron-rich dienophiles. Our experience in the field of dearomatizing cycloadditions will be helpful to tackle this substantial challenge. This [4+2] dearomatizing cycloaddition would lead to a cyclic nitronate that could be subsequently functionalized. The chemistry of cyclic nitronates is a challenging field but offers a large panel of possible C-C bond formation. If the basic principles of their reactivity as nucleophiles are known on 'classical' negatively charged nitronate anions, generated by deprotonation of nitroalkanes, (Henry, Michael reactions) the chemistry of neutral nitronates remains unknown. Activation of the nucleophilic properties of the nitronate may thus be required to get reactive species. In this context, the "umpolung" type reactivity of the structurally related nitrones, which has been described when using samarium diiodide, will probably reveal pivotal. An enantioselective version of this process, catalyzed by chiral acids, or chiral organocatalysts will be envisaged in the course of this study. This approach would be particularly appealing since to date no example of an enantioselective [4+2] heterocycloaddition process involving nitro alkenes as dienes has been described. Alternatively, a diastereoselective approach using chiral enol ethers will be envisaged. Reductive cleavage of the N-O bonds of the compounds thereby produced would then lead to aminated compounds bearing a quaternary center adjacent to the nitrogen atom. Such structural motifs are of difficult access by other synthetic methods. In particular, the direct Csp3-H bond activation adjacent to nitrogen followed by C-C bond formation has been reported in the literature, but remains limited so far. Our method thus bears the advantage of involving trivial starting materials to lead, in two or three synthetic steps only, to dearomatized aminated polycycles featuring a quaternary center at ring junction. The structural analogies of the target polyfunctionalized compounds with many alkaloids make these derivatives potential candidates for biological screening. In addition, the rigid diamino moiety of these scaffolds make them attractive for the design and preparation of novel catalysts in the context of asymmetric synthesis.

The main goal of this project is the development of a new instrument coupling the capabilities of Atom Probe Tomography (APT) and optical spectroscopy, in particular photoluminescence and absorption spectroscopy. The main ambition is to answer to open questions about the physics of ultrafast laser-matter interaction. Furthermore, we aim to develop an instrument which could represent a paradigm for the next generation of atom probes. Atom probe tomography consists in the controlled field ion emission (or “evaporation”) of the atoms of a tip with a nanometric apex radius, immerged in a high electric field. In the latest generation of instruments, this process is triggered by a femtosecond laser pulse. The evaporated ions are detected by a position- and time-of-flight-sensitive detector, which allows for the reconstruction of the chemical composition of the nano-object in 3D. This technique still presents a certain number of open questions, especially concerning the complex laser-matter interaction mechanisms and the physics of field ion emission under ultrafast laser pulse. The absorption mechanisms in insulators, the relaxation of the absorbed energy following the pulse, the dynamics of surface atoms are some examples of the problems requiring a systematic investigation. Moreover, the laser-matter interaction and the evaporation behavior significantly depend on the shape and composition of the studied object, and this may introduce a certain number of aberrations in the trajectories of evaporated ions, which sometimes limit the accuracy of the reconstruction of the original position of the atoms in the sample. Despite these limitations, APT is a technique for nanoscale analysis reaching even beyond the 3D reconstruction. For these reasons, this project proposes to develop new instrumental tools in order to study more in depth the mechanisms of interaction between a fs laser pulse and a nanometric object placed inside a strong electric field (i.e. a field emission tip). The method is based on: (i) The study of selected model systems such as semiconductor quantum emitters and metal nanoparticles embedded in a dielectric matrix, under the form of field emission tips. (ii) A complementary analysis of the mechanisms of optical absorption and emission, as well as of ion field emission, performed simultaneously on the same nano-object. These goals can only be achieved through a preliminary phase of instrumental development. It is indeed necessary to integrate within an atom probe setup all the elements allowing for the optical absorption and emission spectroscopy of a nanometric tip. The primary interest of this project is fundamental, as it focuses on the study of physical phenomena at the nanoscale in conditions (high electric field, ultrafast laser pulse) which have not been systematically addressed yet. The method proposed for this study, based on the coupling of state-of-the-art optical and structural analytical techniques, has never been proposed by other research teams before. The results we expect to obtain will provide a new insight not only in the domain of field ion emission, but also in the domain of the physics and technology of nanostructured systems such as semiconductor nanowires or metal nanoparticles. Finally, the technical know-how acquired by the team in the instrumental development phase can be valorized in terms of intellectual property through the filing of patents. The achievement of the goals of the project are based on the joint competences of the project coordinator (PC) Lorenzo Rigutti, and of the instrumentation team (ERIS) of the Groupe de Physique des Matériaux of the University of Rouen, which he recently joined (September 2012) as a lecturer.

Previous research has shown that stand density reduction by thinning induce a decline in tree leaf litter quality and decomposability of sessile oak (Quercus petraea). The aim of this research project is to investigate how this plasticity in oak litter traits affect forest ecosystem functioning related to soil nitrogen (N) cycling, along with tree N acquisition and competition with understory plants. Combining an in situ study based on a national-scale experimental network with ex situ experiments and using innovative tools such as isotope labelling and metabarcoding, we will test the two following hypotheses: (i) the decline in leaf litter quality related to plastic responses of trees to stand density reduction induces a slowdown of soil N cycling, i.e. inhibiting N mineralization and promoting the dominance of organic N cycling associated with ectomycorrhizal (ECM) fungi; (ii) this slower soil N cycling promotes the fitness of ECM tree seedlings by improving their N acquisition at the expense of competing understory plants. The overall idea is that this decline in leaf litter quality by plasticity could be an adaptive strategy of ECM tree species consisting in short-circuiting nutrient cycling through plant-litter-soil feedbacks. This would preserve their pre-empted nutrient pool of the capture by competing decomposer microbes and understory plants in the context of forest regeneration occurring in treefall gaps or low stand density. Our project will be one of the first attempts to study the importance of phenotypic plasticity in litter quality for biotic interactions, including both plant-soil and plant-plant interactions. It will also allow to better understand the consequences of forest management adaptation to climate change (e.g. stand density reduction by thinning intensification to enhance stand resistance to drought) for the functioning and natural regeneration of forest ecosystems.